<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[AWS Machine Learning for busy people - Medium]]></title>
        <description><![CDATA[AWS Machine Learning certificate exam resources for when you only have time for facts. - Medium]]></description>
        <link>https://medium.com/aws-machine-learning-for-busy-people?source=rss----3e32b8da2ff4---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>AWS Machine Learning for busy people - Medium</title>
            <link>https://medium.com/aws-machine-learning-for-busy-people?source=rss----3e32b8da2ff4---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 22:45:55 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/aws-machine-learning-for-busy-people" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Supervised Learning for Machine Learning]]></title>
            <link>https://medium.com/aws-machine-learning-for-busy-people/supervised-learning-for-machine-learning-22974cdab462?source=rss----3e32b8da2ff4---4</link>
            <guid isPermaLink="false">https://medium.com/p/22974cdab462</guid>
            <category><![CDATA[classification]]></category>
            <category><![CDATA[mls-c01]]></category>
            <category><![CDATA[aws-certification]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[supervised-learning]]></category>
            <dc:creator><![CDATA[Michael Stainsbury]]></dc:creator>
            <pubDate>Thu, 03 Aug 2023 17:12:00 GMT</pubDate>
            <atom:updated>2023-08-03T17:10:47.465Z</atom:updated>
            <content:encoded><![CDATA[<h4>Working with labeled data</h4><figure><img alt="A photo of a grandmother reading aloud to two boys to symbolise supervised learning." src="https://cdn-images-1.medium.com/max/1024/1*VocABbzm4DD0W9YNkZf8WQ.jpeg" /><figcaption>Image by <a href="https://pixabay.com/users/dassel-989431/?utm_source=link-attribution&amp;utm_medium=referral&amp;utm_campaign=image&amp;utm_content=796374">Aline Dassel</a> from <a href="https://pixabay.com/?utm_source=link-attribution&amp;utm_medium=referral&amp;utm_campaign=image&amp;utm_content=796374">Pixabay</a></figcaption></figure><h3>What is Supervised Learning?</h3><p>For Supervised Learning you need labeled training data. In Supervised Learning we provide data that has already been identified, and therefore labeled, as being what we are looking for. Once the Machine Learning model has been trained it can then be presented with real unknown data to which the Machine Learning Model can select a correct label. It is the presence of labels that identify an algorithm as using Supervised Learning</p><p>This article contains revision notes for the AWS certified exam MLS-C01, <em>Machine Learning — Specialty</em>.</p><h3>How is Supervised Learning used?</h3><p>You could, for example, train an algorithm to recognise what healthy fruit looked like and the appearance diseased of diseased fruit. This could be achieved by using training data with pictures of healthy and diseased fruit that had been labelled manually by people. People will have to be trained to identify diseased fruit and then apply the correct labels. To help with this task Amazon provide two services:</p><p>Once the model is trained it can be given images of fruit to determine if they are healthy or diseased.</p><h3>The advantages and disadvantages of Supervised Learning?</h3><p>Supervised learning allows you to predict future events from accumulated historical data. Performance criteria can be optimised using experience. It is usually less computationally intensive since the models can be simpler. The training often requires significant resources due to the quantity of data. The selection of good quality training data is important for performance. However this can lead to the introduction of bias and the risk of over fitting. It is a popular method of solving real world problems.</p><h3>The applications of Supervised Learning?</h3><p>Here are some examples of the applications of Supervised Learning:</p><ul><li>Image recognition</li><li>Predictive analytics. This helps businesses to forecast future events based on the accumulated historical data.</li><li>Customer sentiment analysis. This is the study of text communication from customers to determine their sentiment from the words and phrases they use. Amazon Comprehend is an AI service that also provides sentiment analysis</li><li>Spam detection. Spam emails are detected and directed to a spam directory based on actions performed on similar emails in the past.</li></ul><p>There are two main types of Supervised Learning:</p><ol><li>Classification</li><li>Regression</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/450/1*HOdK4skldAw6bmR8Jy33EA.jpeg" /><figcaption>infographic by Michael Stainsbury</figcaption></figure><h4>Video: Classification and Regression in Machine Learning</h4><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FTJveOYsK6MY%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DTJveOYsK6MY&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FTJveOYsK6MY%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/4edfb61154a26be8e2edc48025e02773/href">https://medium.com/media/4edfb61154a26be8e2edc48025e02773/href</a></iframe><p>This is a brief introduction of Supervised Learning by Max Margenot from Quantopian. It is 2.48 minutes long.</p><h3>Classification Machine Learning</h3><h3>What is Classification?</h3><p>Classification is a type of Supervised Learning that results in data receiving a specific label to identify it as being a member of a class. There are two types of classification:</p><ol><li>Binary classification — two choices or labels</li><li>Multiclass classification — more than two labels</li></ol><p>In Binary Classification there are two choices or labels. An example of binary classification is a model that can detect the presence or absence of disease in a type of fruit. In Multiclass Classification there are more than two choices or labels. An example of a multiclass classification model is one that can identify the specific disease, or disease type.</p><h3>Examples of classes used in Classification</h3><p>These are some examples of classes that can be used with a Classification model. This group are for binary classification where the data is classified in to one of two classes:</p><ul><li>Identification of Male or Female responders to comments on a e-commerce website.</li><li>classification of spam email and non spam email</li><li>positive and negative sentiment in text messages or twitter feeds</li></ul><p>This group are for multi-class classification where the data is classified in to more than two classes:</p><ul><li>classification of types of soil</li><li>classification of types of crops</li><li>classification of mood/feelings in songs/music</li></ul><h3>Applications of Classification Machine Learning</h3><p>Some applications of Classification are:</p><ul><li>speech recognition</li><li>handwriting recognition</li><li>biometric identification</li><li>document classification</li></ul><h3>What SageMaker algorithms use Classification?</h3><ul><li><a href="https://www.mlexam.com/linear-learner-algorithm/">Linear Learner</a> — for binary classification</li><li><a href="https://www.mlexam.com/xgboost-algorithm/">XG Boost</a> — for multiclass classification</li><li><a href="https://www.mlexam.com/k-nearest-neighbors-algorithm/">K Nearest Neighbor</a> — for multiclass classification</li></ul><h3>Recommendation systems</h3><p>Recommendation is a type of Supervised Learning built on top of classification. Recommendations are provided by many of the web based services we interact with, for example:</p><p>Netflix analyzes your previous viewing history to predict films that you are likely to want to watch. This is a Supervised Machine Learning process. All the films have associated metadata, information that describes the film. This may include genre, Age rating, starring actors etc. These are multiclass classifications that are used by the ML algorithms to match, score and rank films to suggest to their customers. Amazon Personalize is AI service that provides personalized recommendations.</p><h4>What SageMaker algorithms support recommendations?</h4><ul><li><a href="https://www.mlexam.com/factorization-machines-algorithm/">Factorization Machines</a></li></ul><h3>Regression for Machine Learning</h3><p>Regression is a supervised learning technique which helps in finding the correlation between variables and enables us to predict the continuous output variable based on the one or more predictor variables.</p><p>A regression problem is when the output variable is a real value such as dollars or weight. The model attempts to find a relationship between dependent and independent variables. The dependent variable is the output. The independent variable is the input. Therefore the output depends on the independent input. These types of values are described as continuous variables because they can be as low as zero, or as high as infinity and assume any value in between. In Regression negative values are often changed to be positive ones to prevent negative predictions. For example temperature in Fahrenheit, which can contain negative values, could be converted to the Kelvin range which only contains positive values.</p><p>There are many types of regression, for example:</p><ul><li>Linear</li><li>Polynomial</li><li>Multiple</li><li>Logistic</li></ul><h4>Video: Making Friends with Machine Learning: Regression</h4><p>This is a 19 minute video by Cassie Kozyrkov from Google. If you need a gentle high level introduction to Regression this is a good place to start.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FWNvOtwP_yf4%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DWNvOtwP_yf4&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FWNvOtwP_yf4%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/eeabcc1a209c83d95ad0fd80b80e7d92/href">https://medium.com/media/eeabcc1a209c83d95ad0fd80b80e7d92/href</a></iframe><h3>Linear Regression</h3><p>Linear regression is a common type of regression used in Machine Learning. The Linear Regression algorithm works by finding the line that is closest to the data points on average. Once this line in known predictions can be made by finding the output from a supplied input using the line. More complex relationships, or graph shapes, can be described by using other Regression techniques.</p><ul><li><a href="https://machinelearningmastery.com/linear-regression-for-machine-learning/">Linear Regression for Machine Learning</a></li><li><a href="https://www.knowledgehut.com/blog/data-science/linear-regression-for-machine-learning">What is Linear Regression?</a></li></ul><h3>What is Time Series Forecasting?</h3><p>Time series analysis is built on top of Regression. Time series data uses time to add order and structure to the data. So the observations are dependent on the time. Time series analysis uses historical data to predict future events. For example:</p><ul><li>Forecasting crop yields</li><li>Forecasting commodities such as gasoline prices</li><li>Stock Market forecasting</li></ul><figure><img alt="A photo of a strawberry spraying machine spraying strawberry plants in a poly tunnel table top system used to illustrate regression type of Unsupervised Machine Learning" src="https://cdn-images-1.medium.com/max/1024/1*uePCOsjNARBTuQCxibhBcw.jpeg" /><figcaption>Image of strawberry sprayer, copyright 2021 <a href="https://npseymour.co.uk/products/table-top-strawberry-sprayer/">NJ Seymore Ltd.</a></figcaption></figure><p>This is an example of regression and time series data. In this case the time period is in years. The line was created by using the Trend Line feature of Microsoft Excel. The data came from a UK government website providing historical data on agriculture. The graph shows the increasing weight of pesticides in strawberry cultivation over time.</p><figure><img alt="An image of a graph showing pesticide use in Strawberry cultivation.. The x axis is time in years and the y axis Kg of pesticide per hectare. A line created from Regression shows the increase of pesticides over time." src="https://cdn-images-1.medium.com/max/481/1*npYhxo-moH9Vbw-90f_zWw.jpeg" /><figcaption>Graph of UK Strawberry pesticide use by Michael Stainsbury</figcaption></figure><p>The equation of the line on the graph is like this:</p><p><strong>y = ax + b</strong></p><p>This simplicity begins to explain why Supervised Learning may require large resources in the training phase, to create the equation, followed by smaller resources in production. This also explains how a Machine Learning model can be installed on edge devices that have constrained resources.</p><h4>What SageMaker algorithms support Regression?</h4><ul><li><a href="https://www.mlexam.com/linear-learner-algorithm/">Linear Learner</a></li><li><a href="https://www.mlexam.com/factorization-machines-algorithm/">Factorization Machines</a></li><li><a href="https://www.mlexam.com/k-nearest-neighbors-algorithm/">K-Nearest Neighbors (k-NN)</a></li><li><a href="https://www.mlexam.com/xgboost-algorithm/">XGBoost</a></li></ul><h3>Summary</h3><p>Supervised Learning is Machine Learning with labeled data. The supervision is the labels and the assessment in training that measures inferences against the known outcome. There are two main types of Supervised Learning: Classification and Regression.</p><p>If you are reading these notes in order the previous revision notes were <a href="https://www.mlexam.com/problem-framing-machine-learning/">Problem Framing for Machine Learning</a>. The next revision notes, when they are published, will be about <a href="https://www.mlexam.com/unsupervised-learning/">Unsupervised Learning</a>.</p><h3>Resources</h3><p>These revision notes are part of sub-domain 3.1, Frame business problems as machine learning problems, of the Modeling domain of the AWS Machine Learning Speciality exam. A description of all the knowledge domains in the exam is in these revision notes: <a href="https://www.mlexam.com/aws-machine-learning-exam-syllabus/">AWS Machine Learning exam syllabus</a></p><ul><li><a href="https://www.guru99.com/supervised-machine-learning.html">Supervised Machine Learning: What is, Algorithms with Examples</a></li><li><a href="https://medium.datadriveninvestor.com/classification-algorithms-in-machine-learning-85c0ab65ff4">Classification Algorithms in Machine Learning…</a></li><li><a href="https://developers.google.com/machine-learning/recommendation">Google course, Recommendation systems</a></li><li><a href="https://machinelearningmastery.com/time-series-forecasting">What Is Time Series Forecasting?</a></li><li><a href="https://www.online-tech-tips.com/ms-office-tips/add-a-linear-regression-trendline-to-an-excel-scatter-plot/">Add a Linear Regression Trendline to an Excel Scatter Plot (online-tech-tips.com)</a></li><li><a href="https://www.gov.uk/government/statistical-data-sets/agriculture-in-the-united-kingdom">Agriculture in the United Kingdom data sets</a></li></ul><h4>Credits</h4><ul><li>Image by <a href="https://pixabay.com/users/dassel-989431/?utm_source=link-attribution&amp;utm_medium=referral&amp;utm_campaign=image&amp;utm_content=796374">Aline Dassel</a> from <a href="https://pixabay.com/?utm_source=link-attribution&amp;utm_medium=referral&amp;utm_campaign=image&amp;utm_content=796374">Pixabay</a></li><li>Image of strawberry sprayer, copyright 2021 <a href="https://npseymour.co.uk/products/table-top-strawberry-sprayer/">NJ Seymore Ltd.</a></li><li>Graph of UK Strawberry pesticide use by Michael Stainsbury</li></ul><p>Unsupervised learning infographic by Michael Stainsbury:</p><ul><li>Image by <a href="https://pixabay.com/users/dassel-989431/?utm_source=link-attribution&amp;utm_medium=referral&amp;utm_campaign=image&amp;utm_content=796374">Aline Dassel</a> from <a href="https://pixabay.com/?utm_source=link-attribution&amp;utm_medium=referral&amp;utm_campaign=image&amp;utm_content=796374">Pixabay</a></li><li>Pickaxe by AB from the <a href="https://thenounproject.com/search/?q=ore&amp;i=254912">Noun Project</a></li><li>golds by iconsphere from the <a href="https://thenounproject.com/search/?q=ore&amp;i=403893">Noun Project</a></li><li>Coins by Phatarawalan Kuapornkiettikul from the <a href="https://thenounproject.com/search/?q=coin&amp;i=2028778">Noun Project</a></li><li><em>Originally published at </em><a href="https://www.mlexam.com/supervised-learning-for-machine-learning/"><em>https://www.mlexam.com</em></a><em> on January 5, 2023.</em></li><li>Copyright <a href="https://www.linkedin.com/in/michael-stainsbury-b695392b/">Michael Stainsbury</a> 2020, 2023</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=22974cdab462" width="1" height="1" alt=""><hr><p><a href="https://medium.com/aws-machine-learning-for-busy-people/supervised-learning-for-machine-learning-22974cdab462">Supervised Learning for Machine Learning</a> was originally published in <a href="https://medium.com/aws-machine-learning-for-busy-people">AWS Machine Learning for busy people</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Problem Framing for Machine Learning]]></title>
            <link>https://medium.com/aws-machine-learning-for-busy-people/problem-framing-for-machine-learning-c1919d1a436a?source=rss----3e32b8da2ff4---4</link>
            <guid isPermaLink="false">https://medium.com/p/c1919d1a436a</guid>
            <category><![CDATA[aws-certification]]></category>
            <category><![CDATA[aws-machine-learning-exam]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[mls-c01]]></category>
            <category><![CDATA[problem-framing]]></category>
            <dc:creator><![CDATA[Michael Stainsbury]]></dc:creator>
            <pubDate>Thu, 03 Aug 2023 17:01:34 GMT</pubDate>
            <atom:updated>2023-08-03T17:05:02.663Z</atom:updated>
            <content:encoded><![CDATA[<h4>subtitle</h4><figure><img alt="A photo of pineapples and a drawing of a pineapple in a frame to symbolise framing a Machine Learning problem." src="https://cdn-images-1.medium.com/max/1024/1*86N3D06iSoylWZlqitn60g.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@pineapple?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Pineapple Supply Co.</a> on <a href="https://unsplash.com/photos/6JNt0yQi37g?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>Problem Framing is a method used to understand, define and prioritize business problems. It is one of the most important phases in Machine Learning that will determine if all the work that is to be done subsequently is perceived to be of use and provides business value. Framing determines what will be observed and what will be predicted. It also identifies the metrics that need to be optimised to monitor performance and errors.</p><p>Problem Framing is preceded by Business Goal Identification. This will provide a clear idea of the problem to be solved and the business value to be derived. This is not a unique ML process, it has been adopted by ML to enable business problems to be phrased so they are suitable for Machine Learning solutions. Framing presents a problem in a way to win support from stakeholders, whilst maximising the chances of a ML solution being successful.</p><p>This article contains revision notes for the AWS certified exam MLS-C01, <em>Machine Learning — Specialty</em>.</p><p>In <em>How To Frame A Problem To Find The Right Solution</em> Paloma Cantero-Gomez offers techniques to help in Problem Framing:</p><p>Forbes: <a href="https://www.forbes.com/sites/palomacanterogomez/2019/04/10/how-to-frame-a-problem-to-find-the-right-solution/?sh=5d25cca45993">How To Frame A Problem To Find The Right Solution</a></p><ul><li>The 40–20–10–5 rule to sequentially cut out words to reach a concise description.</li><li>Rephrase and focus this mostly reframes in a more strategic way.</li><li>Challenge assumptions — what do we know to be true and what have we assumed to be true.</li><li>Change the perspective of the problem by using people from outside of the business area to comment on the problem and so gain a new perspective.</li></ul><p>The Google course <em>Introduction to Machine Learning Problem Framing</em> provides a suggested approach to Problem Framing:</p><p><a href="https://developers.google.com/machine-learning/problem-framing/formulate">Formulate Your Problem as an ML Problem bookmark_border</a></p><ol><li>Articulate your problem.</li><li>Start simple.</li><li>Identify Your Data Sources.</li><li>Design your data for the model.</li><li>Determine where data comes from.</li><li>Determine easily obtained inputs.</li><li>Ability to Learn.</li><li>Think About Potential Bias.</li></ol><h4>Video: Coursera: Lesson 3: Framing Design Problems:</h4><p><a href="https://www.coursera.org/lecture/ux-design-concept-wireframe/lesson-3-framing-design-problems-gZW3u?utm_source=link&amp;utm_medium=page_share&amp;utm_content=vlp&amp;utm_campaign=top_button">Lesson 3: Framing Design Problems — Introduction to the Design Process</a></p><p>You may have to register with Coursera to view this video. Registering is free and you are not asked for payment information.</p><p>The Machine Learning Lens AWS Well-Architected Framework white paper lists six points of best practice:</p><ol><li>Define success criteria</li><li>Establish a performance metric</li><li>Define inputs, outputs and metrics</li><li>Is Machine Learning suitable?</li><li>Data sourcing and annotation objectives</li><li>Select a simple model</li></ol><h3>Problem framing best practice</h3><p>The Machine Learning Lens AWS Well-Architected Framework white paper lists six points of best practice:</p><ol><li>Define success criteria</li><li>Establish a performance metric</li><li>Define inputs, outputs and metrics</li><li>Is Machine Learning suitable?</li><li>Data sourcing and annotation objectives</li><li>Select a simple model</li></ol><h3>Define success criteria</h3><p>Define criteria for a successful outcome of the project. This comes from defining the Business goal.</p><h3>Establish a performance metric</h3><p>Establish an observable and quantifiable performance metric for the project, for example:</p><ul><li>accuracy</li><li>prediction latency</li><li>minimizing inventory value</li></ul><p>This will be used to assess if the training was successful and, when in production, allow the model to be monitored.</p><h3>Inputs, outputs, and performance metrics</h3><p>Formulate the ML question in terms of inputs, desired outputs, and the performance metric to be optimized.</p><h3>Is Machine Learning suitable?</h3><p>Evaluate whether ML is a feasible and appropriate approach. Machine Learning inference works best with large amounts of data that cannot be easily understood, or valuable data points to be identified. The analysis performed so far may indicate that the objectives can be achieved without a Machine Learning process. For example by applying a set of rules, or heuristics.</p><h3>Create a data sourcing and annotation objective</h3><p>Create a data sourcing and data annotation objective, and a strategy to achieve it. Identify where the data will come from and how clean it will be. This will allow you to estimate how much effort will be required to set up a ML Pipeline. You may be able to identify time and cost savings by leaving data out. Issues that cause increased effort to process are:</p><ul><li>Data cleansing</li><li>Data conversion</li><li>Complex data ingestion, or transport</li><li>Personally identifiable information (PII)</li><li>Data requiring enhanced security</li><li>Massive data volumes</li><li>The cost to label</li></ul><p>PII and enhanced security are expensive because of regulatory requirements. For labeling costs, some data may already be labelled, or easily labeled. The most expensive labeling is manual labeling.</p><h3>Select a simple model</h3><p>Start with a simple model that is easy to interpret and implement. More complex models are harder and slow to train than simple models. So by using a simpler model training iterations can be faster which makes debugging more manageable. A more complex model can be introduced later if it is justified.</p><h3>Summary</h3><p>Problem framing is one of the most important phases of the Machine Learning process and one that you have to get right. Everything that comes after relies on the information that comes out of this phase. If you get Framing wrong you can end up doing a lot of work that nobody wants and does not deliver business value.</p><h3>Resources</h3><p>This Study Guide is for sub-domain: 3.1, Frame business problems as machine learning problems, of the Modeling domain of the AWS Machine Learning Speciality exam. A description of all the knowledge domains in the exam is in: <a href="https://www.mlexam.com/aws-machine-learning-exam-syllabus/">AWS Machine Learning exam syllabus</a></p><blockquote>3.1 Frame business problems as machine learning problems.  Determine when to use/when not to use ML  Know the difference between supervised and unsupervised learning  Selecting from among classification, regression, forecasting, clustering, recommendation, etc</blockquote><blockquote><em>AWS Certified Machine Learning — Specialty, (MLS-C01) Exam Guide</em></blockquote><p>I recommend you study the Google course <em>Introduction to Machine Learning Problem Framing.</em> Itis very good, free and only takes about an hour to complete: <a href="https://developers.google.com/machine-learning/problem-framing">Introduction to Machine Learning Problem Framing</a></p><ul><li>AWS White Paper in web page form: <a href="https://docs.aws.amazon.com/wellarchitected/latest/machine-learning-lens/ml-problem-framing.html">ML Problem Framing</a></li><li>AWS White Paper, PDF version: well-architected-machine-learning-lens.pdf</li></ul><h4>Credits</h4><ul><li>Photo by <a href="https://unsplash.com/@pineapple?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Pineapple Supply Co.</a> on <a href="https://unsplash.com/s/photos/frame?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></li><li><em>Originally published at </em><a href="https://www.mlexam.com/problem-framing-machine-learning/"><em>https://www.mlexam.com</em></a><em> on January 6, 2023.</em></li><li>Original article: <a href="https://www.mlexam.com/problem-framing-machine-learning/">https://www.mlexam.com/problem-framing-machine-learning/</a></li><li>Copyright <a href="https://www.linkedin.com/in/michael-stainsbury-b695392b/">Michael Stainsbury</a> 2020, 2023</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c1919d1a436a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/aws-machine-learning-for-busy-people/problem-framing-for-machine-learning-c1919d1a436a">Problem Framing for Machine Learning</a> was originally published in <a href="https://medium.com/aws-machine-learning-for-busy-people">AWS Machine Learning for busy people</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Unsupervised Learning for Machine Learning]]></title>
            <link>https://medium.com/aws-machine-learning-for-busy-people/unsupervised-learning-for-machine-learning-2874a6eb0926?source=rss----3e32b8da2ff4---4</link>
            <guid isPermaLink="false">https://medium.com/p/2874a6eb0926</guid>
            <category><![CDATA[unsupervised-learning]]></category>
            <category><![CDATA[mls-c01]]></category>
            <category><![CDATA[reinforcement-learning]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[aws-certification]]></category>
            <dc:creator><![CDATA[Michael Stainsbury]]></dc:creator>
            <pubDate>Mon, 31 Jul 2023 19:05:50 GMT</pubDate>
            <atom:updated>2023-07-24T13:29:50.037Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Image of a child under three years old reading a fruit alphabet book to symbolize Unsupervised Learning" src="https://cdn-images-1.medium.com/max/1024/1*oSMPqgFZ2SRppgKP8EZ0fA.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@phongduong?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Phong Duong</a> on <a href="https://unsplash.com/photos/HRyg1YnwJ1k?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><h3>What is Unsupervised Learning?</h3><p>Unsupervised learning is the machine learning task of inferring a function to describe hidden structure from unlabeled data. Unsupervised Learning is used to infer patterns in unlabeled datasets. The algorithms can detect hidden patterns and data groupings in data without help from humans through labeling. Unsupervised learning is ideal for exploring raw and unknown data.</p><p>SageMaker built-in algorithms for unsupervised learning:</p><ul><li><a href="https://www.mlexam.com/ip-insights-algorithm/">IP Insights</a></li><li><a href="https://www.mlexam.com/k-means-lgorithm/">K-Means</a></li><li><a href="https://www.mlexam.com/object2vec-lgorithm/">Object2Vec</a></li><li><a href="https://www.mlexam.com/principal-component-analysis-algorithm/">Principal Component Analysis (PCA)</a></li><li><a href="https://www.mlexam.com/random-cut-forest-lgorithm/">Random Cut Forest (RCF)</a></li><li><a href="https://www.mlexam.com/latent-dirichlet-allocation-algorithm/">Latent Dirichlet Allocation (LDA) Algorithm</a></li><li><a href="https://www.mlexam.com/neural-topic-model-algorithm/">Neural Topic Model (NTM) Algorithm</a></li><li><a href="https://www.mlexam.com/blazingtext-algorithm/">BlazingText algorithm</a> — both unsupervised and supervised modes</li></ul><p>This article contains revision notes for the AWS certified exam MLS-C01, <em>Machine Learning — Specialty</em>.</p><h3>Uses of Unsupervised learning</h3><ul><li>Exploratory data analysis to visualize patterns in data. Unsupervised learning may be used for data exploration prior to identifying opportunities for Supervised Learning.</li><li>Cross selling marketing strategies</li><li>Customer segmentation</li><li>Image recognition</li></ul><p>Wikipedia: <a href="https://en.wikipedia.org/wiki/Unsupervised_learning">Unsupervised learning</a></p><p>IBM: <a href="https://www.ibm.com/cloud/learn/unsupervised-learning">What is Unsupervised Learning?</a></p><h3>What are the advantages of unsupervised learning?</h3><p>The big advantage of Unsupervised Learning is that it does not need labeled data. Labeling large datasets can be very expensive especially if it requires human interaction to perform the labeling. Labeling is also difficult when the number or identity of the classes may not be known, or may be expected to change over time. This freedom from the burden of labeling means that Unsupervised Learning techniques can be used on new or changing data.</p><ul><li><a href="https://towardsdatascience.com/unsupervised-learning-and-data-clustering-eeecb78b422a">Unsupervised Learning and Data Clustering</a></li><li><a href="https://pythonprogramminglanguage.com/the-importance-of-unsupervised-learning">The importance of unsupervised learning</a></li></ul><h3>What types of Unsupervised Learning are there?</h3><p>Unsupervised Learning techniques and algorithms can be divided into two groups:</p><ol><li>Probabilistic techniques</li><li>Neural Networks</li></ol><h3>Probabilistic Unsupervised Learning</h3><h3>What probabilistic techniques are there in Unsupervised Learning?</h3><p>Probabilistic techniques are statistical methods that are used to examine unlabeled data to find groups and similarities. This also enables data that is dissimilar to be identified. Common techniques for unsupervised learning are:</p><ul><li>Clustering</li><li>Topic modeling</li><li>Embeddings</li><li>Anomaly detection</li><li>Dimensionality Reduction</li></ul><h3>Video: Amazon SageMaker’s Built-in Algorithm Webinar Series: Clustering with K Means</h3><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FRPInYpI9MjY&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DRPInYpI9MjY&amp;image=http%3A%2F%2Fi.ytimg.com%2Fvi%2FRPInYpI9MjY%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/a7e2388b5936d4858f2f726946699951/href">https://medium.com/media/a7e2388b5936d4858f2f726946699951/href</a></iframe><p>This video by Kris Skrinak is quite long, however only the first 8.18 minutes are of specific relevance to this study guide. I recommend you watch the first 8.18 minutes now and come back to it later when you have completed revising the <a href="https://www.mlexam.com/k-means-lgorithm/">SageMaker K-Means built in algorithm</a>.</p><h3>Clustering</h3><ul><li><a href="https://towardsdatascience.com/unsupervised-learning-and-data-clustering-eeecb78b422a">Unsupervised Learning and Data Clustering</a></li></ul><h3>What is Clustering in Unsupervised Learning?</h3><p>Clustering or Cluster Analysis is a group of techniques used to collect objects into groups based on features or attributes, these groups are called Clusters. A Cluster is a group of objects that have features that are similar to each other and dis-similar to the features of other clusters.</p><p>The features that determine if an object is a member of a cluster may be simply present or absent, or they may form part of a spectrum. In this case the members of a Cluster will have features close to each other and distant from members in other Clusters. This can be described as distance or proximity.</p><p>The advantage of Clustering is that it provides an insight to the structure of the data. This can be used to formulate a <a href="https://www.mlexam.com/supervised-learning-for-machine-learning/">Supervised Learning</a> strategy.</p><h4>What types of Clustering are there?</h4><p>There are different types of clustering including:</p><ul><li>Exclusive clustering</li><li>Overlapping clustering</li><li>Hierarchical clustering</li><li>Probabilistic clustering</li></ul><h4>What SageMaker algorithms support Clustering?</h4><ul><li><a href="https://www.mlexam.com/k-means-lgorithm/">K-Means</a></li><li><a href="https://www.mlexam.com/latent-dirichlet-allocation-algorithm/">LDA</a></li></ul><h3>Topic Modeling</h3><ul><li><a href="https://springerplus.springeropen.com/articles/10.1186/s40064-016-3252-8">An overview of topic modeling and its current applications in bioinformatics</a></li><li><a href="https://monkeylearn.com/blog/introduction-to-topic-modeling/">Introduction to Topic Modeling Topic modeling is an unsupervised machine learning technique that’s capable of scanning</a></li></ul><h3>What is Topic Modeling in Unsupervised Learning?</h3><p>Topic modeling is used to analyze text based data such as documents, emails, text messages etc. It uses clustering techniques for discovering topics that occur in a collection of documents. Algorithms are chosen to extract topics that are clear, segregated and meaningful. The output of Topic Modeling is a list of topics, which may be words or phrases, and groups of documents identified as containing topics from the list.</p><h3>What is the advantage of Topic Modeling?</h3><p>The advantage of Topic Modeling is that it does not need labeled data. This means that it can be rapidly deployed on incoming data and change the topics it identifies as the input data changes.</p><h3>What are the applications of Topic Modeling?</h3><p>The applications of Topic Modeling are:</p><ul><li>Sentiment Analysis — Topic Modeling can be used to identify what the users sentiment was from the message they sent. This can be used for marketing feedback to analyse how people felt about a product or service.</li><li>Opinion summarization — This is a type of Sentiment Analysis to relate people’s opinions to specific documented policies, or processes.</li><li>Help desk / call center automation — by using Topic Modeling the topics of the request for assistance can be identified and the call routed to the most appropriate response.</li><li>Chat bots can use Topic Modeling to identify what the user wants the chatbot to do and match this with documents and resources that the user needs.</li><li>Question and answer — Many websites have document search features to encourage user self help. The topics can be extracted from the question and be matched to the appropriate help pages which can be ranked in order of the match to the identified topic.</li><li>Spam filters — Using Topic Modeling text based messaging systems such as email and text messages can be scanned to identify messages likely to be spam by the topics they contain. These can then be filtered out to a spam folder, or removed entirely.</li><li>Bioinformatics — This is an area of current research. Biological research can produce vast quantities of data. Topic Modeling can be used to identify biological <em>topics</em> for further investigation.</li></ul><p><a href="https://medium.com/@fatmafatma/industrial-applications-of-topic-model-100e48a15ce4">Industrial applications of topic model</a></p><h3>What SageMaker algorithms support Topic Modeling?</h3><ul><li><a href="https://www.mlexam.com/latent-dirichlet-allocation-algorithm/">LDA (Latent Dirichlet Allocation)</a></li><li><a href="https://www.mlexam.com/neural-topic-model-algorithm/">NTM (Neural Topic Model)</a></li></ul><h3>Embeddings</h3><h3>What are Embeddings in Unsupervised Learning?</h3><p>Embeddings are a way of learning about an object once and then saving the learning in a form that allows it to be <em>embedded</em> in other learning models. So Embeddings are a way of reusing the results of prior training in other models.</p><ul><li><a href="https://machinelearningmastery.com/what-are-word-embeddings/">What Are Word Embeddings for Text?</a></li></ul><p>The definition of embeddings provided by the Google Machine Learning Crash Course is:</p><blockquote>“An is a relatively low-dimensional space into which you can translate high-dimensional vectors. Embeddings make it easier to do machine learning on large inputs like sparse vectors representing words. Ideally, an embedding captures some of the semantics of the input by placing semantically similar inputs close together in the embedding space. An embedding can be learned and reused across models.”</blockquote><blockquote><em>Machine Learning Crash Course, Google</em></blockquote><p>An example of Embeddings is understanding words and phrases. However the objects for Embeddings are not restricted to only text.</p><h3>What are the advantages of Embeddings?</h3><p>By reusing inferences created by prior learning efforts the Model will work more efficiently than if it had to learn these inferences by itself. Also the Embeddings may be of highly quality if they were created using superior training data.</p><h3>What SageMaker algorithms support Embeddings?</h3><ul><li><a href="https://www.mlexam.com/object2vec-lgorithm/">Object2Vec</a></li></ul><h3>Anomaly Detection</h3><h3>What is Anomaly Detection in Unsupervised Learning?</h3><p>Anomaly detection is used to identify patterns that do not conform to expected or usual behavior. This is defined more specifically in the Google article <em>Use real-time anomaly detection reference patterns to combat fraud</em> as:</p><blockquote><em>“Anomaly detection-or in broader terms, outlier detection-allows businesses to identify and take action on changing user needs, detect and mitigate malignant actors and behaviors, and take preventive actions to reduce costly repairs.”</em></blockquote><blockquote>Use real-time anomaly detection reference patterns to combat fraud<em>, Shan Kulandaivel &amp; Cody Irwin, Google</em></blockquote><ul><li><a href="https://cloud.google.com/blog/products/data-analytics/using-automated-ml-streaming-architecture-to-find-anomalies">Using automated ML, streaming architecture to find anomalies</a></li><li>Oracle: <a href="https://blogs.oracle.com/datascience/introduction-to-anomaly-detection">Introduction to Anomaly Detection</a></li></ul><h3>What is Anomaly Detection used for?</h3><ul><li>Network Traffic monitoring</li><li>Anti fraud</li><li>Security analytics</li><li>Fault detection in operating environments</li></ul><h3>What types of Anomaly are there?</h3><p>Anomalies can be described in three general groups:</p><ul><li>Point Anomalies</li><li>Contextual Anomalies</li><li>Collective Anomalies</li></ul><p>Point Anomalies are where a single instance of data is anomalous if it’s too unlike the rest. An example is detecting financial fraud based on the amount spent in one transaction. In Contextual Anomalies the event could be normal in some circumstances, but not in others. For example high spend on alcohol may usually occur on weekends, but not Monday to Thursday; multiple purchases of high denomination gift vouchers outside the Christmas season. With Collective Anomalies we are looking for a series of events that together are unusual. For example, changes to security permissions, moving large files, downloading to a location outside the organisation may indicate a cyber attack.</p><h3>What is the advantage of Anomaly Detection?</h3><p>The advantage of Anomaly Detection is that it enables the detection of events that have not yet been identified as being detrimental, or advantageous. This means that you can identify events of interest that have never happened before allowing an organisation to take early action. Rapid detection allows for rapid mitigation to reduce the cost of the Anomaly or to take advantage of it. Because the detection is automatic it eliminates the manual burden of identifying unusual activity.</p><h3>What SageMaker algorithms support Anomaly Detection?</h3><ul><li><a href="https://www.mlexam.com/random-cut-forest-lgorithm/">Random Cut Forest</a></li><li><a href="https://www.mlexam.com/ip-insights-algorithm/">IP Insights</a></li></ul><h3>Dimensionality Reduction</h3><h3>What is Dimensionality Reduction in Unsupervised Learning?</h3><p>Dimensionality Reduction is concerned with reducing the number of features. For example by reducing a feature set from fifty to twenty columns. Dimensionality Reduction techniques are used to reduce the dimension of the feature set, so that the new feature set retains some meaningful properties of the original data.</p><p>Dimensionality is discussed in detail in these revision notes: See <a href="https://www.mlexam.com/feature-engineering/">Feature Engineering (Sub-domain 2.2)</a></p><h3>What SageMaker algorithms support Dimensionality Reduction?</h3><ul><li><a href="https://www.mlexam.com/principal-component-analysis-algorithm/">Principle Component Analysis (PCA)</a></li></ul><h3>Unsupervised Learning using Neural Networks</h3><h3>What are Neural Networks?</h3><p>Neural Networks are algorithms that mimic the human brain’s capacity to learn. Neural Networks are made up of layers of neurons, or nodes, that are interlinked in a network. The advantage of Neural Networks is that once they are trained they can make inferences very fast and with great accuracy.</p><ul><li>IBM: <a href="https://www.ibm.com/cloud/learn/neural-networks">What are Neural Networks?</a></li></ul><h3>How do Neural Networks work?</h3><p>Neural Networks are made up of layers of nodes organised into an input layer, an output layer and multiple hidden layers. Each node is linked to all of the nodes in the next layer. A node contains the maths that drives the Neural Networks and determines what happens when the input is received. The outcome will be to pass information on to the next layer, or make no response. This is the part of the Neural Network that can be trained.</p><h3>Video: Neural Networks and Deep Learning: Crash Course AI #3</h3><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FoV3ZY6tJiA0%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DoV3ZY6tJiA0&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FoV3ZY6tJiA0%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/26908e45ee5b169cf7cc93f8af61577d/href">https://medium.com/media/26908e45ee5b169cf7cc93f8af61577d/href</a></iframe><p>A gentle introduction by Jabril from Crash Course AI. This video does not contain much maths. Don’t be put off by the cringy props that wouldn’t amuse an eight year old. This video is worth watching and is only 12.22 minutes long.</p><h3>What types of Neural Networks are there?</h3><p>There are many different types of Neural Networks which differ in three ways:</p><ul><li>The shape, or topology of the network</li><li>The types of node or cells in the network</li><li>The linking between the nodes</li></ul><p>This article has a good image summarising many of them:</p><ul><li><a href="https://towardsdatascience.com/the-mostly-complete-chart-of-neural-networks-explained-3fb6f2367464">The mostly complete chart of Neural Networks, explained</a></li></ul><h3>How does SageMaker support Neural Networks?</h3><p>SageMaker supports Neural Networks with it’s Deep Learning features and services.</p><h3>Deep Learning</h3><ul><li><a href="https://machinelearningmastery.com/what-is-deep-learning/">What is Deep Learning?</a></li><li>AWS: <a href="https://aws.amazon.com/deep-learning/">Get Started with Deep Learning on AWS</a></li></ul><h3>What is Deep Learning in Unsupervised Learning?</h3><p>Deep Learning is another name for large Neural Networks. The depth refers to the number of layers in the network. However there is no agreed definition of what if Deep Learning.</p><h3>What are the advantages of Deep Learning?</h3><p>The advantages of Deep Learning are:</p><ul><li>Speed — Neural Networks are designed for speed.</li><li>Scalability — because Neural Networks are fast they can handle large quantities of unlabeled data.</li><li>Flexibility — There are a range of Deep Learning frameworks and interfaces available for a range of use cases.</li></ul><h3>Example applications of Deep Learning</h3><ul><li>Computer vision</li><li>Speech recognition</li><li>Natural Language Processing</li><li>Recommendation engines</li></ul><p>Deep Learning frameworks and interfaces are discussed further in: <a href="https://www.mlexam.com/aws-ml-services/#Deep-Learning-Frameworks-and-Infrastructures">Machine Learning services and features.</a></p><h3>Reinforcement Learning</h3><ul><li><a href="https://www.mlexam.com/aws-ml-services/#amazon-sagemaker-rl">Amazon SageMaker RL</a></li><li><a href="https://docs.aws.amazon.com/sagemaker/latest/dg/reinforcement-learning.html">Use Reinforcement Learning with Amazon SageMaker</a></li><li><a href="https://www.geeksforgeeks.org/what-is-reinforcement-learning/">Reinforcement learning</a></li></ul><h3>What is Reinforcement Learning in Unsupervised Learning?</h3><p>AWS defines Unsupervised Reinforcement Learning (RL) as:</p><blockquote><em>“In unsupervised learning, an agent learns by uncovering structure within unlabeled data. While a RL agent might benefit from uncovering structure based on its experiences, the sole purpose of RL is to maximize a reward signal.”</em></blockquote><blockquote><em>Use Reinforcement Learning with Amazon SageMaker, Amazon Webservices</em></blockquote><ul><li><a href="https://docs.aws.amazon.com/sagemaker/latest/dg/reinforcement-learning.html">Use Reinforcement Learning with Amazon SageMaker</a></li></ul><p>Reinforcement learning draws it’s heritage from research into the psychology of learning processes of real animals and people. In the 1980s it was combined with computational research to produce the Reinforcement Learning used in Machine Learning today.</p><ul><li>History: <a href="http://incompleteideas.net/book/ebook/node12.html">1.6 History of Reinforcement Learning</a></li></ul><h3>What are the advantages of Reinforcement Learning?</h3><p>The advantages of Reinforcement Learning are:</p><ul><li>Maximizes Performance</li><li>Sustain Change for a long period of time</li><li>Can be used on new or changing data</li></ul><h3>How does Reinforcement Learning work?</h3><p>In Reinforcement Learning an Agent (a program) seeks out answers to problems and receives a reward when the problem is solved. Next time the Agent has the same or similar problem the same solution is tried again and so leads more rapidly to the reward. To do this the Agent has learned how to get the reward. The Agent seeks out actions to receive the reward and stops performing actions that do not lead to the reward. So the learning is reinforced by the reward.</p><p>Most Reinforcement Learning models can be described as Markov Decision Processes (MDP). A Markov Decision Process comprises one or more Episodes. Each Episode is made up of one or more Time Steps. Each Time Step has the following features:</p><ul><li>Environment</li><li>State -This is the current situation and environment which is relevant to what Actions the Agent can take next.</li><li>Action — What the Agent does.</li><li>Reward — This is the Reward that may be received for a successful Action.</li><li>Observation — This is what the Agent can perceive of the Environment</li></ul><h3>How does SageMaker support Reinforcement Learning?</h3><p>In SageMaker, Reinforcement Learning (RL) is implemented using SageMaker supported Deep Learning frameworks, interfaces and toolkits. The supported Deep Learning frameworks are TensorFlow and Apache MxNet. The SageMaker RL Toolkit manages interactions between the Agent and the environment and also provides RL algorithms. SageMaker also supports the Intel Coach and Ray RLlib toolkits. There is a wide range of environments available and SageMaker supports the Open AI Gym Interface.</p><h4>Video: What is Reinforcement Learning? | AI 101</h4><p>This video by Jordan Harrod is 10.47 minutes long.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FO0mVwDuZAtc%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DO0mVwDuZAtc&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FO0mVwDuZAtc%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/e4d78c97563e1393a7fa151abf3b02eb/href">https://medium.com/media/e4d78c97563e1393a7fa151abf3b02eb/href</a></iframe><h3>Summary</h3><p>Unsupervised Learning opens the possibilities of finding out things about your data that you never knew were there. This is due to the potential of unlabeled data where the endgame may not be fully known. The two types of Unsupervised Learning are Probabilistic and Neural Networks. The Probabilistic methods are based on statistics and are supported in SageMaker by SageMaker built in algorithms. Neural Networks are built using SageMaker Deep Learning features and services which include the frameworks TensorFlow and Apache MxNet. SageMaker also supports Reinforcement Learning in this way.</p><h3>Resources</h3><blockquote>Domain 3: Modeling, 3.1 Frame business problems as machine learning problems. Determine when to use/when not to use ML; Know the difference between supervised and unsupervised learning; Selecting from among classification, regression, forecasting, clustering, recommendation, etc.</blockquote><blockquote><em>AWS Certified Machine Learning — Specialty, (MLS-C01) Exam Guide</em></blockquote><p>Unsupervised learning is part of <a href="https://www.mlexam.com/problem-framing-machine-learning/">sub-domain 3.1, Frame business problems as machine learning problems</a>, which is in <a href="https://www.mlexam.com/home/domain-3-modeling/">domain 3, Modeling</a>. A description of all the knowledge domains in the exam is in these revision notes: <a href="https://www.mlexam.com/aws-machine-learning-exam-syllabus/">AWS Machine Learning exam syllabus</a></p><p><a href="https://www.mlexam.com/problem-framing-machine-learning/#test-questions-app">5 questions and answers from a test bank to 10 questions.</a></p><p><a href="https://www.mlexam.com/aws-machine-learning-practice-exam/">Free MLS-C01 AWS Machine Learning practice exam</a></p><p>Resources mentioned in this article:</p><ul><li>Wikipedia: <a href="https://en.wikipedia.org/wiki/Unsupervised_learning">Unsupervised learning</a></li><li>IBM: <a href="https://www.ibm.com/cloud/learn/unsupervised-learning">What is Unsupervised Learning?</a></li><li><a href="https://towardsdatascience.com/unsupervised-learning-and-data-clustering-eeecb78b422a">Unsupervised Learning and Data Clustering</a></li><li><a href="https://pythonprogramminglanguage.com/the-importance-of-unsupervised-learning">The importance of unsupervised learning</a></li><li><a href="https://towardsdatascience.com/unsupervised-learning-and-data-clustering-eeecb78b422a">Unsupervised Learning and Data Clustering</a></li><li><a href="https://springerplus.springeropen.com/articles/10.1186/s40064-016-3252-8">An overview of topic modeling and its current applications in bioinformatics</a></li><li><a href="https://monkeylearn.com/blog/introduction-to-topic-modeling/">Introduction to Topic Modeling Topic modeling is an unsupervised machine learning technique that’s capable of scanning</a></li><li><a href="https://medium.com/@fatmafatma/industrial-applications-of-topic-model-100e48a15ce4">Industrial applications of topic model</a></li><li><a href="https://machinelearningmastery.com/what-are-word-embeddings/">What Are Word Embeddings for Text?</a></li><li>Google Machine Learning crash course video: <a href="https://developers.google.com/machine-learning/crash-course/embeddings/video-lecture">Embeddings</a></li><li><a href="https://cloud.google.com/blog/products/data-analytics/using-automated-ml-streaming-architecture-to-find-anomalies">Using automated ML, streaming architecture to find anomalies</a></li><li>Oracle: <a href="https://blogs.oracle.com/datascience/introduction-to-anomaly-detection">Introduction to Anomaly Detection</a></li><li>IBM: <a href="https://www.ibm.com/cloud/learn/neural-networks">What are Neural Networks?</a></li><li><a href="https://towardsdatascience.com/the-mostly-complete-chart-of-neural-networks-explained-3fb6f2367464">The mostly complete chart of Neural Networks, explained</a></li><li><a href="https://machinelearningmastery.com/what-is-deep-learning/">What is Deep Learning?</a></li><li>AWS: <a href="https://aws.amazon.com/deep-learning/">Get Started with Deep Learning on AWS</a></li><li><a href="https://www.mlexam.com/aws-ml-services/#Deep-Learning-Frameworks-and-Infrastructures">Machine Learning services and features.</a></li><li><a href="https://www.mlexam.com/aws-ml-services/#amazon-sagemaker-rl">Amazon SageMaker RL</a></li><li><a href="https://docs.aws.amazon.com/sagemaker/latest/dg/reinforcement-learning.html">Use Reinforcement Learning with Amazon SageMaker</a></li><li><a href="https://www.geeksforgeeks.org/what-is-reinforcement-learning/">Reinforcement learning</a></li><li><a href="https://docs.aws.amazon.com/sagemaker/latest/dg/reinforcement-learning.html">Use Reinforcement Learning with Amazon SageMaker</a></li><li>History: <a href="http://incompleteideas.net/book/ebook/node12.html">1.6 History of Reinforcement Learning</a></li></ul><h3>Credits</h3><p>Photo by<a href="https://unsplash.com/@phongduong?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Phong Duong</a> on<a href="https://unsplash.com/?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Unsplash</a></p><p><em>Originally published at </em><a href="https://www.mlexam.com/unsupervised-learning/"><em>https://www.mlexam.com</em></a><em> on January 23, 2023.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2874a6eb0926" width="1" height="1" alt=""><hr><p><a href="https://medium.com/aws-machine-learning-for-busy-people/unsupervised-learning-for-machine-learning-2874a6eb0926">Unsupervised Learning for Machine Learning</a> was originally published in <a href="https://medium.com/aws-machine-learning-for-busy-people">AWS Machine Learning for busy people</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to select a model for a given machine learning problem]]></title>
            <link>https://medium.com/aws-machine-learning-for-busy-people/how-to-select-a-model-for-a-given-machine-learning-problem-62dc95743e9c?source=rss----3e32b8da2ff4---4</link>
            <guid isPermaLink="false">https://medium.com/p/62dc95743e9c</guid>
            <category><![CDATA[sagemaker]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[mls-c01]]></category>
            <category><![CDATA[aws-certification]]></category>
            <category><![CDATA[select-a-model]]></category>
            <dc:creator><![CDATA[Michael Stainsbury]]></dc:creator>
            <pubDate>Mon, 31 Jul 2023 19:04:55 GMT</pubDate>
            <atom:updated>2023-07-26T05:02:05.543Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8fcG-bAkwTwnAclbw4culQ.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@dari_lli?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Dari lli</a> on <a href="https://unsplash.com/@dari_lli?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>Amazon SageMaker has over 17 built-in algorithms. But how do you choose which one to use? This selection process is the next step after <a href="https://www.mlexam.com/problem-framing-machine-learning/">Framing the Problem</a>. The information and conclusions from problem framing tells us the type of problem to be solved and the data which will be used for Model training and inference.</p><p>This article covers the subject material for<a href="https://www.mlexam.com/aws-machine-learning-exam-guide/"> subdomain 3,2 of the AWS certified exam MLS-C01, <em>Machine Learning — Specialty</em></a>.</p><p>A Machine Learning problem can be described with four aspects:</p><ol><li>Data types and format</li><li>Learning paradigm or domain</li><li>Problem type</li><li>Use case examples</li></ol><p>The first aspect concerns the format and structure of the data, which could be numeric, images or text. Numeric data is often tabular. The second aspect is the learning paradigm which includes supervised learning, unsupervised learning. The third aspect is about the type of problem, for example classification, clustering, or topic modeling. The final aspect is use cases and AWS provides seventeen use case examples which will apply to many Machine Learning problems.</p><p>This information allows us to narrow down the choices of algorithms, sometimes to a single algorithm. However there may be other factors that influence the choice of algorithm. For example, some algorithms do not perform well with sparse data. These factors and nuances are discussed in the individual algorithm pages.</p><h4>Video: Built-in Machine Learning Algorithms with Amazon SageMaker — a Deep Dive</h4><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FyGc0qePSYig&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DyGc0qePSYig&amp;image=http%3A%2F%2Fi.ytimg.com%2Fvi%2FyGc0qePSYig%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/d4ff7becadc4011a4b9cee138970feae/href">https://medium.com/media/d4ff7becadc4011a4b9cee138970feae/href</a></iframe><p>To select a model SageMaker has seventeen built in algorithms to choose from. These are optimised versions of common open source algorithms. Here they are listed in alphabetical order:</p><ol><li><a href="https://www.mlexam.com/blazingtext-algorithm/">BlazingText algorithm</a></li><li><a href="https://www.mlexam.com/deepar-forecasting-algorithm/">DeepAR Forecasting Algorithm</a></li><li><a href="https://www.mlexam.com/factorization-machines-algorithm/">Factorization Machines Algorithm</a></li><li><a href="https://www.mlexam.com/image-classification-algorithm/">Image Classification Algorithm</a></li><li><a href="https://www.mlexam.com/ip-insights-algorithm/">IP Insights</a></li><li><a href="https://www.mlexam.com/k-means-lgorithm/">K-Means Algorithm</a></li><li><a href="https://www.mlexam.com/k-nearest-neighbors-algorithm/">K-Nearest Neighbors (K-NN) Algorithm</a></li><li><a href="https://www.mlexam.com/latent-dirichlet-allocation-algorithm/">Latent Dirichlet Allocation (LDA) Algorithm</a></li><li><a href="https://www.mlexam.com/linear-learner-algorithm/">Linear Learner Algorithm</a></li><li><a href="https://www.mlexam.com/neural-topic-model-algorithm/">Neural Topic Model (NTM) Algorithm</a></li><li><a href="https://www.mlexam.com/object-detection-algorithm/">Object Detection Algorithm</a></li><li><a href="https://www.mlexam.com/object2vec-lgorithm/">Object2Vec Algorithm</a></li><li><a href="https://www.mlexam.com/principal-component-analysis-algorithm/">Principal Component Analysis (PCA) Algorithm</a></li><li><a href="https://www.mlexam.com/random-cut-forest-lgorithm/">Random Cut Forest (RCF) Algorithm</a></li><li><a href="https://www.mlexam.com/semantic-segmentation-algorithm/">Semantic Segmentation Algorithm</a></li><li><a href="https://www.mlexam.com/sequence-to-sequence-algorithm/">Sequence-to-Sequence Algorithm</a></li><li><a href="https://www.mlexam.com/xgboost-algorithm/">XGBoost Algorithm</a></li></ol><h4>Video: AWS re:Invent 2020: Choose the right machine learning algorithm in Amazon SageMaker</h4><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FY8FzvaQBS6s%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DY8FzvaQBS6s&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FY8FzvaQBS6s%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/db022928004ec94b3bc3b933ad64a0eb/href">https://medium.com/media/db022928004ec94b3bc3b933ad64a0eb/href</a></iframe><p>This is a 29.53 minutes video from AWS by Denis Batalov and Alberto Danese. The timestamps are:</p><ul><li>0 — introduction</li><li>1.30–17 built-in algorithms in Amazon SageMaker</li><li>5.13 — Image classification demo</li><li>10.35 — Guide for classification / regression algorithm</li><li>11.30 — Amazon blog post on Linear Learner</li><li>11.46 — K-Nearest Neighbor (K-NN)</li><li>13.22 — Amazon blog post on K-NN</li><li>13.50 — XG Boost, how it works</li><li>16.00 — Getting a grasp on how XG Boost works</li><li>24.00 — XG Boost as a built-in algorithm</li><li>25.18 — XG Boost in Nexi</li><li>28.19 — Popular frameworks</li><li>28.47 — AWS Marketplace</li><li>29.20 — Amazon resources for SageMaker built-in algorithms</li></ul><h4>What is the definition of a model and an algorithm</h4><p>When you take one of the SageMaker built-in algorithms and train it with data you create a model, therefore:</p><p><strong>Model = Training (an Algorithm + Data)</strong></p><p>(See: <a href="https://medium.com/@kaykwaalk/difference-between-an-algorithm-and-a-model-in-machine-learning-5d6b7297cce9">difference-between-an-algorithm-and-a-model-in-machine-learning</a>)</p><h3>The four aspects of a problem used for model selection</h3><ol><li>Data types and format</li><li>Learning paradigm or domain</li><li>Problem type</li><li>Use case examples</li></ol><h3>Data types and format</h3><p>SageMaker algorithms have very specific requirements for the data you train them with. So we can select a model based on the data type and form of the data the algorithm processes. This aspect allows the SageMaker built in algorithms to be split into three groups: Tabular, Text and Image data.</p><p><strong>Tabular data</strong></p><p>SageMaker built in algorithms that use numeric data in a tabular format as the input data structure.</p><ul><li><a href="https://www.mlexam.com/deepar-forecasting-algorithm/">DeepAR Forecasting</a></li><li><a href="https://www.mlexam.com/factorization-machines-algorithm/">Factorization Machines</a></li><li><a href="https://www.mlexam.com/ip-insights-algorithm/">IP Insights</a></li><li><a href="https://www.mlexam.com/k-means-lgorithm/">K-Means</a></li><li><a href="https://www.mlexam.com/k-nearest-neighbors-algorithm/">K-Nearest Neighbors (K-NN)</a></li><li><a href="https://www.mlexam.com/linear-learner-algorithm/">Linear Learner</a></li><li><a href="https://www.mlexam.com/object2vec-lgorithm/">Object2Vec</a></li><li><a href="https://www.mlexam.com/principal-component-analysis-algorithm/">Principal Component Analysis (PCA)</a></li><li><a href="https://www.mlexam.com/random-cut-forest-lgorithm/">Random Cut Forest (RCF)</a></li><li><a href="https://www.mlexam.com/xgboost-algorithm/">XGBoost</a></li></ul><p><strong>Text data</strong></p><p>SageMaker built in algorithms that use text as the input data.</p><ul><li><a href="https://www.mlexam.com/blazingtext-algorithm/">BlazingText</a></li><li><a href="https://www.mlexam.com/latent-dirichlet-allocation-algorithm/">Latent Dirichlet Allocation (LDA)</a></li><li><a href="https://www.mlexam.com/neural-topic-model-algorithm/">Neural Topic Model (NTM)</a></li><li><a href="https://www.mlexam.com/sequence-to-sequence-algorithm/">Sequence-to-Sequence</a></li></ul><p><strong>Image data</strong></p><p>SageMaker built in algorithms that use images as the input data.</p><ul><li><a href="https://www.mlexam.com/image-classification-algorithm/">Image Classification</a></li><li><a href="https://www.mlexam.com/object-detection-algorithm/">Object Detection</a></li><li><a href="https://www.mlexam.com/semantic-segmentation-algorithm/">Semantic Segmentation</a></li></ul><h3>Learning paradigm</h3><p>The learning paradigms are:</p><ol><li>Supervised learning</li><li>Unsupervised learning</li></ol><p>The Learning Paradigm narrows the search for a suitable algorithm. The selection of which paradigm to use depends on if the data is labelled or unlabelled. <a href="https://www.mlexam.com/supervised-learning-for-machine-learning/">Supervised Learning</a> requires labeled data, <a href="https://www.mlexam.com/unsupervised-learning/">Unsupervised Learning</a> can use unlabeled data.</p><p>All three algorithms that use image data require labeled data and so use supervised learning.</p><ul><li><a href="https://www.mlexam.com/image-classification-algorithm/">Image Classification</a></li><li><a href="https://www.mlexam.com/object-detection-algorithm/">Object Detection</a></li><li><a href="https://www.mlexam.com/semantic-segmentation-algorithm/">Semantic Segmentation</a></li></ul><p>Of the four algorithms that use text data two, LDA and NTM, can use unlabeled data and so perform unsupervised learning. One algorithm, Sequence-to-Sequence, uses labeled data for supervised learning. BlazingText is the only algorithm that has modes for both supervised and unsupervised learning.</p><ul><li><a href="https://www.mlexam.com/blazingtext-algorithm/">BlazingText algorithm</a> — both</li><li><a href="https://www.mlexam.com/latent-dirichlet-allocation-algorithm/">Latent Dirichlet Allocation (LDA) Algorithm</a> — unsupervised</li><li><a href="https://www.mlexam.com/neural-topic-model-algorithm/">Neural Topic Model (NTM) Algorithm</a> — unsupervised</li><li><a href="https://www.mlexam.com/sequence-to-sequence-algorithm/">Sequence-to-Sequence Algorithm</a> — supervised</li></ul><p>Of the remaining ten algorithms half use supervised learning and half unsupervised learning. These algorithms use numeric tabular data.</p><p><strong>Supervised learning</strong></p><ul><li><a href="https://www.mlexam.com/deepar-forecasting-algorithm/">DeepAR Forecasting</a></li><li><a href="https://www.mlexam.com/sagemaker-supervised-algorithms/#factorization-machines">Factorization Machines</a></li><li><a href="https://www.mlexam.com/k-nearest-neighbors-algorithm/">K-Nearest Neighbors (K-NN)</a></li><li><a href="https://www.mlexam.com/linear-learner-algorithm/">Linear Learner</a></li><li><a href="https://www.mlexam.com/xgboost-algorithm/">XGBoost</a></li></ul><p><strong>Unsupervised learning</strong></p><ul><li><a href="https://www.mlexam.com/ip-insights-algorithm/">IP Insights</a></li><li><a href="https://www.mlexam.com/k-means-lgorithm/">K-Means</a></li><li><a href="https://www.mlexam.com/object2vec-lgorithm/">Object2Vec</a></li><li><a href="https://www.mlexam.com/principal-component-analysis-algorithm/">Principal Component Analysis (PCA)</a></li><li><a href="https://www.mlexam.com/random-cut-forest-lgorithm/">Random Cut Forest (RCF)</a></li></ul><h3>Problem type</h3><p>Algorithms are designed to give answers to known problems. These problems can be grouped into general Problem Types. The Problem Type is the type of problem with reference to the data. The Problem Type aspect includes:</p><ul><li>Classification<br>- <a href="https://www.mlexam.com/factorization-machines-algorithm/">Factorization Machines</a><br>- <a href="https://www.mlexam.com/k-nearest-neighbors-algorithm/">K-Nearest Neighbors (k-NN)</a><br>- <a href="https://www.mlexam.com/xgboost-algorithm/">XGBoost</a></li><li>Regression<br>- <a href="https://www.mlexam.com/factorization-machines-algorithm/">Factorization Machines</a><br>- <a href="https://www.mlexam.com/k-nearest-neighbors-algorithm/">K-Nearest Neighbors (k-NN)</a><br>- <a href="https://www.mlexam.com/linear-learner-algorithm/">Linear Learner</a><br>- <a href="https://www.mlexam.com/xgboost-algorithm/">XGBoost</a></li><li>Time-series forecasting<br>- <a href="https://www.mlexam.com/deepar-forecasting-algorithm/">DeepAR Forecasting</a><br>- <a href="https://www.mlexam.com/linear-learner-algorithm/">Linear Learner</a></li><li>Clustering<br>- <a href="https://www.mlexam.com/k-means-lgorithm/">K-Means</a></li><li>Topic modeling<br>- <a href="https://www.mlexam.com/latent-dirichlet-allocation-algorithm/">Latent Dirichlet Allocation (LDA)</a><br>- <a href="https://www.mlexam.com/neural-topic-model-algorithm/">Neural Topic Model (NTM)</a></li><li>Dimensionality reduction<br>- <a href="https://www.mlexam.com/principal-component-analysis-algorithm/">Principal Component Analysis (PCA)</a></li><li>Anomaly detection<br>- <a href="https://www.mlexam.com/random-cut-forest-lgorithm/">Random Cut Forest (RCF) Algorithm</a><br>- <a href="https://www.mlexam.com/ip-insights-algorithm/">IP Insights</a></li><li>Embeddings<br>- <a href="https://www.mlexam.com/object2vec-lgorithm/">Object2Vec</a><br>- <a href="https://www.mlexam.com/blazingtext-algorithm/">BlazingText</a></li><li>Text classification<br>- <a href="https://www.mlexam.com/blazingtext-algorithm/">BlazingText</a></li><li>Machine translation<br>- <a href="https://www.mlexam.com/sequence-to-sequence-algorithm/">Sequence-to-Sequence</a></li><li>Image and multi-label classification<br>- <a href="https://www.mlexam.com/image-classification-algorithm/">Image Classification Algorithm</a></li><li>Object detection and classification<br>- <a href="https://www.mlexam.com/object-detection-algorithm/">Object Detection</a></li><li>Computer vision<br>- <a href="https://www.mlexam.com/semantic-segmentation-algorithm/">Semantic Segmentation</a></li></ul><h3>Use case examples</h3><p>SageMaker built-in algorithms are designed with use cases in mind. Typically there is only one algorithm for a use case, but sometimes there are more than one. When there are multiple use cases the algorithms are differentiated by other features and properties.</p><p><strong>Numeric use cases</strong></p><ul><li><a href="https://www.mlexam.com/deepar-forecasting-algorithm/">DeepAR Forecasting</a> — Based on historical data for a behavior, predict future behavior</li><li><a href="https://www.mlexam.com/linear-learner-algorithm/">Linear Learner </a>— Predict a numeric/continuous value; Predict if an item belongs to a category</li><li><a href="https://www.mlexam.com/factorization-machines-algorithm/">Factorization Machines</a> — Predict a numeric/continuous value; Predict if an item belongs to a category</li><li><a href="https://www.mlexam.com/k-nearest-neighbors-algorithm/">K-Nearest Neighbors (K-NN)</a> — Predict a numeric/continuous value; Predict if an item belongs to a category</li><li><a href="https://www.mlexam.com/xgboost-algorithm/">XGBoost</a> — Predict a numeric/continuous value; Predict if an item belongs to a category</li><li><a href="https://www.mlexam.com/random-cut-forest-lgorithm/">Random Cut Forest (RCF)</a> — Detect abnormal behavior in application</li><li><a href="https://www.mlexam.com/ip-insights-algorithm/">IP Insights</a> — Protect your application from suspicious users</li><li><a href="https://www.mlexam.com/k-means-lgorithm/">K-Means</a> — Group similar objects/data together</li><li><a href="https://www.mlexam.com/principal-component-analysis-algorithm/">Principal Component Analysis (PCA)</a> — Drop those columns from a dataset that have a weak relation with the label/target variable. This reduces the number of features to be analyzed.</li><li><a href="https://www.mlexam.com/object2vec-lgorithm/">Object2Vec</a> — Improve the data embeddings of the high-dimensional objects</li></ul><p><strong>Text use cases</strong></p><ul><li><a href="https://www.mlexam.com/sequence-to-sequence-algorithm/">Sequence-to-Sequence</a> — Convert audio files to text, Summarize a long text corpus, Convert text from one language to others</li><li><a href="https://www.mlexam.com/blazingtext-algorithm/">BlazingText</a> — Assign predefined categories to documents in a corpus of text</li><li><a href="http://nmt/">Neural Topic Model (NTM)</a> — Organize a set of documents into topics (not known in advance)</li><li><a href="https://www.mlexam.com/latent-dirichlet-allocation-algorithm/">Latent Dirichlet Allocation (LDA)</a> — Organize a set of documents into topics (not known in advance)</li></ul><p><strong>Image use cases</strong></p><ul><li><a href="https://www.mlexam.com/image-classification-algorithm/">Image Classification</a> — Label/tag an image based on the content of the image</li><li><a href="https://www.mlexam.com/object-detection-algorithm/">Object Detection</a> — Detect people and objects in an image</li><li><a href="https://www.mlexam.com/semantic-segmentation-algorithm/">Semantic Segmentation</a> — Tag every pixel of an image individually with a category</li></ul><h3>Classifying algorithms with Learning Paradigm and Data Type</h3><p>The first two problem aspects we discussed have the fewest options, two for learning paradigms, Supervised and Unsupervised and three for data types, tabular numeric, text and images. These aspects are also the easiest to identify in Problem Framing since they are based on easily observable characteristics of the data. From the table below it can be seen that determining if the data will require Supervised or Unsupervised learning will make a significant reduction in the number of possible algorithms to choose from. However you will still have up to five algorithms to choose from each group.</p><h3>Conclusion</h3><p>SageMaker has seventeen built-in algorithms that can be used to build Machine Learning models. Four aspects can be used to select a model: Data types and format; Learning paradigm or domain; Problem type; Use case examples. Using these aspects to select appropriate algorithms will reduce choice to a small group and often to a single algorithm.</p><h3>Resources</h3><ul><li>AWS: <a href="https://docs.aws.amazon.com/sagemaker/latest/dg/algos.html">https://docs.aws.amazon.com/sagemaker/latest/dg/algos.html</a></li><li>AWS choose algorithm: <a href="https://docs.aws.amazon.com/sagemaker/latest/dg/algorithms-choose.html">https://docs.aws.amazon.com/sagemaker/latest/dg/algorithms-choose.html</a></li><li>AWS: <a href="https://docs.aws.amazon.com/sagemaker/latest/dg/algos.html">https://docs.aws.amazon.com/sagemaker/latest/dg/algos.html</a></li><li><a href="https://www.kdnuggets.com/2020/05/guide-choose-right-machine-learning-algorithm.html">https://www.kdnuggets.com/2020/05/guide-choose-right-machine-learning-algorithm.html</a></li><li><a href="https://towardsdatascience.com/do-you-know-how-to-choose-the-right-machine-learning-algorithm-among-7-different-types-295d0b0c7f60">https://towardsdatascience.com/do-you-know-how-to-choose-the-right-machine-learning-algorithm-among-7-different-types-295d0b0c7f60</a></li></ul><p>These revision notes are part of subdomain 3.2 <em>Select the appropriate model(s) for a given machine learning problem</em> of the <a href="https://www.mlexam.com/aws-machine-learning-exam-syllabus/">exam guide.</a></p><blockquote>3.2 Select the appropriate model(s) for a given machine learning problem.  Xgboost, logistic regression, K-means, linear regression, decision trees, random forests, RNN, CNN, Ensemble, Transfer learning</blockquote><blockquote>Express intuition behind models</blockquote><blockquote><em>AWS Certified Machine Learning — Specialty, (MLS-C01) Exam Guide</em></blockquote><p>The original article contains a questions and answers test app that asks five questions from a test bank of ten questions: <a href="https://www.mlexam.com/how-to-select-a-model-for-a-given-machine-learning-problem#test-questions-app">Visit ML Exam website</a></p><p>More questions for SageMaker built-in algorithms and their uses are in this article: <a href="https://www.mlexam.com/35-q-a-for-sagemaker-built-in-algorithms/">35 Q &amp; A for SageMaker built-in algorithms</a></p><h4>Credits</h4><p>Photo by<a href="https://unsplash.com/@dari_lli?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Dari lli</a> on<a href="https://unsplash.com/s/photos/cookie-cutters?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Unsplash</a></p><p><em>Originally published at </em><a href="https://www.mlexam.com/how-to-select-a-model-for-a-given-machine-learning-problem/"><em>https://www.mlexam.com</em></a><em> on January 5, 2023.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=62dc95743e9c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/aws-machine-learning-for-busy-people/how-to-select-a-model-for-a-given-machine-learning-problem-62dc95743e9c">How to select a model for a given machine learning problem</a> was originally published in <a href="https://medium.com/aws-machine-learning-for-busy-people">AWS Machine Learning for busy people</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AWS Machine Learning exam guide]]></title>
            <link>https://medium.com/aws-machine-learning-for-busy-people/aws-machine-learning-exam-guide-8a3f9c5879af?source=rss----3e32b8da2ff4---4</link>
            <guid isPermaLink="false">https://medium.com/p/8a3f9c5879af</guid>
            <category><![CDATA[aws-certification]]></category>
            <category><![CDATA[mls-c01]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[exam-guide]]></category>
            <dc:creator><![CDATA[Michael Stainsbury]]></dc:creator>
            <pubDate>Tue, 18 Jul 2023 15:41:39 GMT</pubDate>
            <atom:updated>2023-08-03T16:52:22.955Z</atom:updated>
            <content:encoded><![CDATA[<h4>A guide to the exam guide</h4><figure><img alt="A photo of a lady reading a map with the road disappearing in the distance symbolising the AWS Machine Learning certificate exam guide." src="https://cdn-images-1.medium.com/max/1024/1*pByxDHbyt-zDTRtj_wyzWw.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@overlyawesome?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Daniel Gonzalez</a> on <a href="https://unsplash.com/s/photos/map?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>Syllabus, specification and blue print are all terms to describe the knowledge domain of an exam or course. However AWS call the description of the content of their Machine Learning exam the <em>Exam Guide</em>. Perhaps this is a telling choice since the information they provide is far from comprehensive, it is just a guide.</p><p>If you come from an AWS Machine Learning background the Exam Guide PDF will be sufficient for you. However if you are already a Data Scientist and wish to move into AWS, or you use AWS and want to learn to use SageMaker and Machine Learning then large chunks of the Exam Guide will be unintelligible. This <em>guide to the exam guide</em> fills the gaps and explains high level concepts.</p><p>The Exam Guide is where the exam subjects are listed, split into four domains and fifteen sub-domains. This article describes each sub-domain in enough detail for the complete newbie to get a good idea of what it is about. If you intend to study for the AWS Machine Learning certificate this will give you an overview of what you are getting yourself into.</p><ul><li>AWS pdf: <a href="https://d1.awsstatic.com/training-and-certification/docs-ml/AWS-Certified-Machine-Learning-Specialty_Exam-Guide.pdf">https://d1.awsstatic.com/training-and-certification/docs-ml/AWS-Certified-Machine-Learning-Specialty_Exam-Guide.pdf</a></li></ul><h3>Domain 1: Data Engineering</h3><p><a href="https://www.mlexam.com/home/domain-1-data-engineering/">Domain 1 Data Engineering</a> is concerned with obtaining the data, transforming it and putting it in a repository. It comprises 20% of the exam marks. There are three sub-domains that can be summarised as:</p><ul><li><a href="https://www.mlexam.com/data-repositories/">1.1 Data repositories</a></li><li><a href="https://www.mlexam.com/data-cleansing/">1.2 Data ingestion</a></li><li><a href="https://www.mlexam.com/data-transformation/">1.3 Data transformation</a></li></ul><h4>1.1 Data repositories</h4><p><em>Create data repositories for machine learning</em></p><p>The data repository is where raw and processed data is stored. S3 is the repository of choice for Machine Learning in AWS and all built-in algorithms and services can consume data from S3. Other data stores are also mentioned in the exam guide:</p><ul><li>Database (Relational Database Service)</li><li>Data Lake (LakeFormation)</li><li>EFS</li><li>EBS</li></ul><p>Often data is generated by the business themselves, but sometimes data from other sources is needed to train the model. For example libraries of image data to train the Object Detection algorithm. Many data sources are publicly available.</p><h4>1.2 Data ingestion</h4><p><em>Identify and implement a data ingestion solution</em></p><p>The data ingestion sub-domain is concerned with gathering the raw data into the repository. This can be via batch processing or streaming data. With <a href="https://www.mlexam.com/batch-processing/">batch processing</a>, data is collected and grouped at a point in time and passed to the data store. <a href="https://www.mlexam.com/streaming-data-for-machine-learning/">Streaming data</a> is constantly being collected and fed into the data store. The AWS streaming services are:</p><ul><li>Kinesis family of streaming data services:<br>- Kinesis Data Streams<br>- Kinesis Firehose<br>- Kinesis Analytics<br>- Kinesis Video Streams</li></ul><figure><img alt="An infographic describing the Kinesis family of streaming data services: Kinesis Data Streams, Kinesis Firehose, Kinesis Analytics and Kinesis Video Streams" src="https://cdn-images-1.medium.com/max/450/1*sWzQBUYaBiyvrRT8BVT2UQ.jpeg" /><figcaption>Infographic by Author</figcaption></figure><p>Batch processing requires a way to schedule or trigger the processing, also called job scheduling. Examples are Glue Workflow and Step Functions. AWS batch services include:</p><ul><li>EMR (Hadoop)</li><li>Glue (Spark)</li></ul><h4>1.3 Data transformation</h4><p><em>Identify and implement a data transformation solution</em></p><p>The third Data Engineering sub-domain focuses on how raw data is transformed into data that can be used for ML processing. The transformation process changes the data structure. The data may also need to be clean up, de-duplicated, incomplete data managed and have it’s attributes standardised. The AWS Services are similar to those used with data ingestion:</p><ul><li>Glue (Spark)</li><li>EMR (Hadoop, Spark, Hive)</li><li>AWS Batch</li></ul><p>Once these data engineering processes are complete the data is ready for further pre-processing prior to being fed into a Machine Learning algorithm. This preprocessing is covered by the second knowledge domain <a href="https://www.mlexam.com/home/domain-2-exploratory-data-analysis/">Exploratory Data Analysis</a>.</p><h3>Domain 2: Exploratory Data Analysis</h3><p>In the <a href="https://www.mlexam.com/home/domain-2-exploratory-data-analysis/">Exploratory Data Analysis</a> domain the data is analysed so it can be understood and cleaned up. It comprises 24% of the exam marks. There are three sub-domains:</p><ul><li><a href="https://www.mlexam.com/data-cleansing/">2.1 Prep and sanitise data</a></li><li><a href="https://www.mlexam.com/feature-engineering/">2.2 Feature engineering</a></li><li><a href="https://www.mlexam.com/data-visualization/">2.3 Analyse and visualize data</a></li></ul><h4>2.1 Prep and sanitise data</h4><p><em>Sanitize and prepare data for modeling</em></p><p>In <a href="https://www.mlexam.com/data-cleansing/"><em>Sanitize and prepare data for modeling</em></a>, the data can be cleaned up using techniques to remove distortions and fill in gaps.</p><ul><li>missing data</li><li>corrupt data</li><li>stop words</li><li>formatting</li><li>normalizing</li><li>augmenting</li><li>scaling data</li></ul><figure><img alt="An infographic showing techniques used to clean data." src="https://cdn-images-1.medium.com/max/450/1*Nvq3J8nTD6FMC9zXo3EuJw.jpeg" /><figcaption>Infographic by Author</figcaption></figure><p><a href="https://aws.amazon.com/sagemaker/data-labeling/what-is-data-labeling/">Data labeling is the process of identifying raw data and adding one or more meaningful and informative labels to provide context.</a> Data labeling can be costly and time consuming because it involves applying the labels manually. AWS provides the service called Mechanical Turk to reduce the cost and speed up the labelling process.</p><h4>2.2 Feature engineering</h4><p><em>Perform feature engineering</em></p><p><a href="https://www.mlexam.com/feature-engineering/">Feature Engineering</a> is about creating new features from existing ones to make the Machine Learning algorithms more powerful. Feature Engineering techniques are used to reduce the number of features and categorise the data.</p><ul><li>binning</li><li>tokenization</li><li>outliers</li><li>synthetic features</li><li>1 hot encoding</li><li>reducing dimensionality of data</li></ul><figure><img alt="An infographic showing aspects of feature engineering." src="https://cdn-images-1.medium.com/max/800/1*J73CtblhxMCIzuM9x-4aew.jpeg" /><figcaption>Infographic by Author</figcaption></figure><h4>2.3 Analyse and visualize data</h4><p><em>Analyze and visualize data for machine learning</em></p><p><a href="https://www.mlexam.com/data-visualization/">Analyzing and visualizing the data</a> overlaps with the other two sub-domains which use these techniques. The techniques include graphs, charts and matrices.</p><ul><li>scatter plot</li><li>histogram</li><li>box plot</li></ul><p>Before data can be sanitized and prepared is has to bet understood. This is done using statistics that focus on specific aspects of the data and graphs and charts that allow relationships and distributions to be seen.</p><ul><li>correlation</li><li>summary statistics</li><li>p value</li><li>elbow plot</li><li>cluster size</li></ul><p>You now understand your data and have cleaned it up ready for the next stage, modeling.</p><h3>Domain 3: Modeling</h3><p>When people talk about Machine Learning they are mostly thinking about Modeling. Modeling is selecting and testing the algorithms to process data to find the information of value. It comprises 36% of the exam marks. This domain has five sub-domains:</p><ul><li><a href="https://www.mlexam.com/problem-framing-machine-learning/">3.1 Frame the business problem</a></li><li><a href="https://www.mlexam.com/how-to-select-a-model-for-a-given-machine-learning-problem/">3.2 Select the appropriate model</a></li><li><a href="https://www.mlexam.com/training-machine-learning-models/">3.3 Train the model</a></li><li><a href="https://www.mlexam.com/model-tuning/">3.4 Tune the model</a></li><li><a href="https://www.mlexam.com/how-to-evaluate-machine-learning-models/">3.5 Evaluate the model</a></li></ul><h4>3.1 Frame the business problem</h4><p><em>Frame business problems as machine learning problems</em></p><p>First we decide if Machine Learning is appropriate for this problem. Machine Learning is good for data driven problems involving large amounts of data where the rules cannot easily be coded. The business problem can probably be framed in many ways and this determines what kind of Machine Learning problem is being solved. For example the business problem could be framed to require a yes/no answer as in fraud detection, or a value as in share price. Also in this sub-domain we can find out the type of data and so if the algorithm will use a supervised, or unsupervised paradigm. From the type of problem that has to be solved features of the algorithm can be identified, for example classification, regression, forecasting, clustering or recommendation.</p><figure><img alt="An infographic describing best practice when framing Machine Learning problems." src="https://cdn-images-1.medium.com/max/450/1*h9rLboijdNqSPV3oPNfmWQ.jpeg" /><figcaption>Infographic by Author</figcaption></figure><h4>3.2 Select the appropriate model</h4><p><em>Select the appropriate model(s) for a given machine learning problem</em></p><p>Many models are available through AWS Machine Learning services with SageMaker having over seventeen built-in algorithms. Each model has it’s own use cases and requirements. Once the model has been chosen an iterative process of training, tuning and evaluation is undertaken.</p><p>The exam guide lists the SageMaker built-in algorithms XGboost, K-means. Since there are many built-in algorithms perhaps these are just the most important. Modeling concepts are also listed:</p><ul><li>linear regression — Linear Learner, K-Nearest Neighbors, Factorization Machines</li><li>logistic regression — XGBoost</li><li>decision trees — XGBoost</li><li>random forests — Random Cut Forest</li><li>RNN — DeepAR forecasting, Sequence to Sequence</li><li>CNN — Sequence to Sequence</li><li>ensemble learning — XGBoost</li><li>transfer learning — Image classification</li></ul><h4>3.3 Train the model</h4><p><em>Train machine learning models</em></p><p>Model training is the process of providing a model with data to learn from. During model training the data is split into three parts. Most is used as training data with the remainder used for validation and testing. Cross validation is a technique used when training data is limited. By understanding the concepts of the internal workings of algorithms, model training can be optimised. Concepts used by models in training include gradient descent, loss functions, local minima, convergence, batches, optimizer and probability.</p><p>The speed and cost of training depends on the choices about the compute resources used. The type of instance central processing unit can be specified. Graphical Processing Units (GPU) can provide more compute power, but not all algorithms can utilise them and may require cheaper CPU instances. For heavy training loads distributed processing options may be available to speed up training. Spark and non-Spark data processing can be used to pre-process training data.</p><p>Model training is also concerned with how and when models are updated and retrained.</p><h4>3.4 Tune the model</h4><p><em>Perform hyperparameter optimization</em></p><p>Model tuning is also known as hyperparameter optimisation. Machine Learning algorithms can be thought of as black boxes with hyperparameters being the exposed controls that can be changed and optimised. Hyperparameters settings do not change during training. They can be tuned manually before training commences, using search methods or automatically by using SageMaker guided search. Model tuning can be improved by using:</p><ul><li>Regularization</li><li>Drop out</li><li>L1/L2</li><li>Model initialization</li></ul><p>Models that utilise a neural network architecture use other hyperparameters:</p><ul><li>layers / nodes</li><li>learning rate</li><li>activation functions</li></ul><p>Tree-based models have hyperparameters that influence the number of trees and number of levels. The learning rate is used to optimise Linear models.</p><h4>3.5 Evaluate the model</h4><p><em>Evaluate machine learning models</em></p><p>Model evaluation is used to find out how well a Model will perform in predicting the desired outcome. This is done using metrics to measure the performance of the Model. Metrics measure accuracy, precision and other features of the Model by comparing the results from the Model with the known contents of the training data.</p><p>Metrics commonly used:</p><ul><li>AUC-ROC</li><li>accuracy</li><li>precision</li><li>recall</li><li>RMSE</li><li>F1 score</li></ul><p>To measure the correlation between features a Confusion matrix is used.</p><p>Evaluation methods can be performed offline or online. A/B testing can also be used to compare the performance of model variants. Metrics allow the detecting of a poorly fitting model caused by bias or variance. This is where a model performs poorly with real world data.<br><br>Other metrics allow models and model variants to be compared using metrics that are not directly related to data:</p><ul><li>time to train a model</li><li>quality of model</li><li>engineering costs</li></ul><p>Cross validation</p><p>Your model is now ready to be used with real data. But before it can be let loose on your corporate data it has to be deployed into the production environment.</p><h3>Domain 4: Machine Learning Implementation and Operations</h3><p>This domain is about Systems Architecture and DevOps skills to make everything work in production. It comprises 20% of the exam marks. There are four sub-domains:</p><ul><li><a href="https://www.mlexam.com/machine-learning-production-environment/">4.1 Build machine learning solutions for performance, availability, scalability, resiliency, and fault tolerance.</a></li><li><a href="https://www.mlexam.com/aws-ml-services/">4.2 Recommend and implement the appropriate machine learning services and features for a given problem.</a></li><li><a href="https://www.mlexam.com/aws-security/">4.3 Apply basic AWS security practices to machine learning solutions.</a></li><li><a href="https://www.mlexam.com/deploy-ml-model/">4.4 Deploy and operationalize machine learning solutions.</a></li></ul><h4>4.1 The production environment</h4><p><em>Build machine learning solutions for performance, availability, scalability, resiliency, and fault tolerance.</em></p><p>Designing AWS production environments for performance, availability, scalability, resiliency, and fault tolerance is part of AWS best practice. Resilience and availability is provided by deploying models in multiple AWS regions and multiple Availability Zones. Auto Scaling groups and Load balancing provide scalability for compute resources. Performance is optimised by rightsizing EC2 instances, volumes and provision IOPS. There are a variety of deployment options including EC2, SageMaker managed EC2 via endpoints and docker containers. CloudTrail and CloudWatch are used for AWS environment logging and monitoring. This assists in creating fault tolerance systems and build error monitoring.</p><h4>4.2 ML services and features</h4><p><em>Recommend and implement the appropriate machine learning services and features for a given problem</em></p><p>AWS provides a range of services and features to choose from for a given Machine Learning problem. AWS provide AI services which are highly optimised algorithms deployed on AWS managed infrastructure. Some of the services contain pre-trained models ready for production inferencing. Some examples are:</p><ul><li>Poly, text to speech</li><li>Lex, chatbot</li><li>Transcribe, speech to text</li></ul><figure><img alt="An Infographic listing the Amazon AI services provided by AWS." src="https://cdn-images-1.medium.com/max/450/1*zO7feQJ3WCIBGlk_f-Bzkg.jpeg" /><figcaption>Infographic by Author</figcaption></figure><p>When using AI services AWS does all the heavy lifting of managing infrastructure, models and training. There are other features if you need to have more control of these aspects. SageMaker built-in algorithms can be used or you can bring your own model. This allows cost considerations to influence the choice of compute services. Even more sophisticated cost control can be achieved with using spot instances to train deep learning models using AWS Batch</p><p>AWS Service limits are used to limit the amount of resources. This may be a limit on the number of instances of a service used in an account. AWS Service limits can be increased by AWS on request. Sometimes there is a hard limit which is the maximum for that service in a single AWS Account or Region.</p><h4>4.3 Security</h4><p><em>Apply basic AWS security practices to machine learning solutions</em></p><p>Security in AWS starts with the ubiquitous IAM, Identity and Access Management, which controls the activities of all AWS services. Since S3 is the most common storage for Machine Learning services S3 bucket policies are also included. It may seem that access to VPCs, <a href="https://docs.aws.amazon.com/vpc/latest/userguide/what-is-amazon-vpc.html">Amazon Virtual Private Cloud</a>, and VPC <a href="https://docs.aws.amazon.com/vpc/latest/userguide/vpc-security-groups.html">Security Groups</a> may not be needed if you are implementing serverless applications. However, under the hood, SageMaker uses these services and the security has to be configured. As well as configuring security for the services, data security also has to be considered. This includes encryption of data both at rest and in transit. Anonymisation can be used to protect PII data, <a href="https://docs.aws.amazon.com/comprehend/latest/dg/pii.html">Personally identifiable information</a>.</p><h4>4.4 Deploy and operationalize</h4><p><em>Deploy and operationalize machine learning solutions</em></p><p>There are many ways to deploy Machine Learning models in production, one method is to use SageMaker endpoints. Despite the name a SageMaker endpoint is more than an isolated interface, it sits on top of serious processing power. This is provided by SageMaker managed EC2 instances which are set up by the endpoint configuration. SageMaker endpoints can host multiple variants of the same model. This enables different variants of the same Model to be compared using testing strategies, for example A/B testing.</p><p>Once in production the model is monitored because the performance of a model may degrade over time as real world data changes. This drop in performance can be detected and used to trigger the retraining of the model via a retrain pipeline.</p><h3>Summary</h3><p>The AWS Certified Machine Learning — Speciality, exam guide is good for outlining the breadth of the course and how it is divided up into four domains and fifteen sub-domains. Whilst it lists and mentions many subjects only a few are described in any detail and it is still a bit light with those. This article provides additional description of the subjects to allow someone considering studying for the exam to understand what has to be learnt to achieve exam success.</p><h4>Credits</h4><ul><li>Photo by <a href="https://unsplash.com/@overlyawesome?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Daniel Gonzalez</a> on <a href="https://unsplash.com/s/photos/map?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></li><li><em>Originally published at </em><a href="https://www.mlexam.com/aws-machine-learning-exam-guide/"><em>https://www.mlexam.com</em></a><em> on October 8, 2020.</em></li><li>Copyright <a href="https://www.linkedin.com/in/michael-stainsbury-b695392b/">Michael Stainsbury</a> 2020, 2023</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8a3f9c5879af" width="1" height="1" alt=""><hr><p><a href="https://medium.com/aws-machine-learning-for-busy-people/aws-machine-learning-exam-guide-8a3f9c5879af">AWS Machine Learning exam guide</a> was originally published in <a href="https://medium.com/aws-machine-learning-for-busy-people">AWS Machine Learning for busy people</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Free questions for the AWS Machine Learning exam]]></title>
            <link>https://medium.com/aws-machine-learning-for-busy-people/free-questions-for-the-aws-machine-learning-exam-4fdb9e468908?source=rss----3e32b8da2ff4---4</link>
            <guid isPermaLink="false">https://medium.com/p/4fdb9e468908</guid>
            <category><![CDATA[free]]></category>
            <category><![CDATA[mls-c01]]></category>
            <category><![CDATA[aws-certification]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[free-practice-test]]></category>
            <dc:creator><![CDATA[Michael Stainsbury]]></dc:creator>
            <pubDate>Tue, 18 Jul 2023 15:40:35 GMT</pubDate>
            <atom:updated>2023-07-12T06:02:19.403Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="A photo of a forest with question marks spray painted on to the tree trunks to symbolise Machine Learning questions and exams" src="https://cdn-images-1.medium.com/max/1024/1*w1qXBllkTtxupXIQJMtxvw.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@evan__bray?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Evan Dennis</a> on <a href="https://unsplash.com/s/photos/questions?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>Exam style questions are an important resource to pass the AWS Machine Learning Speciality certificate exam. My strategy is to binge study exam style questions as the last phase of study before taking the exam. By the time I come to exam style questions I have already studied the subject areas I need to understand and practiced with labs and tutorials. This is the last stage before the real exam. I use the scores from the questions to check I am good enough to pass the exam. At $300 this is no time for a near miss. The pass mark is about 75% so once my scores are regularly in the mid eighties I know I am ready and I book the exam.</p><p>The exam comprises approximately 65 questions that have to be answered in 180 minutes. The pass mark varies from 75% to 80%. Extra time is allowed for people with English as a second language and other special needs such as dyslexia. To get the extra time you must book the exam with a special title and code, not the standard exam. The special exam may have rules and tasks that have to be completed to allow you to book it, so you need to research this before you need to book the exam.</p><p>Note that whilst the full price for the exam is $300 if you have passed an AWS certificate in the past you may have a 50% voucher assigned to you. Go to your AWS training account and search the benefits section. There you will find images of exam logo badges and any money off vouchers.</p><p>The exam contains four knowledge domains each contributing a different percentage of the exam. For more information about the exam see these revision notes: <a href="https://www.mlexam.com/aws-machine-learning-exam-syllabus/">AWS Machine Learning Exam Guide</a></p><p>Each question consists of a text based question followed by multiple answers. Some questions require more than one answer to be chosen. These typically follow the usual format of answers you did at school with one crazy answer that can be rejected immediately and two or three very similar answers that you must choose from. There are no marks deducted for incorrect answers so it is a good policy to guess if you don’t know the answer. The computer app that delivers the questions has a feature that lets you mark questions to come back to later.</p><p>When writing this guide to free questions I searched for <em>free questions for the AWS Machine Learning exam</em> on Google and found a large group of vendors offering free questions, usually as a sample of their paid for offering. However before I recommended them to you I wanted to vet them to make sure I was not using my reputation to back up a vendor that did not deserve it.</p><h4>Checking the vendors</h4><p>I assessed the vendors using this criteria:</p><ol><li>Their website offered free exam questions without asking you to sign up for anything or hand over your email address.</li><li>Their website looked as though there was ‘life’ behind it. That someone was actually home.</li><li>The person or company behind the website could be identified on the About page, or Terms and Conditions and could be contacted by more than just a web form.</li><li>The website did not look scammy or fraudulent.</li><li>The website had acceptable reviews on TrustPilot.</li></ol><p>I started with eleven results from the first two pages of the Google search copied into a Google Doc and then applied my criteria. To my amazement most failed the test.</p><h4>Ghosts and secrets</h4><p>Some sites were <em>ghost</em> sites with the latest blog posts a year old and in one case no evidence of activity for three years. Most of those that failed were anonymous sites. Sites that had no contact details and gave no clue as to the people or organisation behind them. The websites were sophisticated, well maintained e-commerce sites. So this was not an innocent oversight this was a policy to evade responsibility for what they were doing.</p><h4>Costly reader software</h4><p>One site sold a downloadable file of questions for $30 that were impossible to read without special reader software that cost much more than the original purchase. Whilst the web page did mention that the reader was needed, it did not make it clear that a subsequent purchase was necessary.</p><h4>Trustpilot reviews</h4><p>TrustPilot: <a href="https://www.trustpilot.com/">https://www.trustpilot.com/</a></p><p>Whilst some of these issues were forgivable it was the TrustPilot reviews that were most damning. All businesses get some bad reviews and some of the reviewers were clearly complaining about something that was not the vendor’s fault, such as not passing the exam. However multiple complaints about answers to questions being incorrect shows that there was a quality problem. Reading between the lines some of the positive reviews appeared to have been coerced from customers seeking a refund. In another case the vendor appears to have been submitting their own reviews. All of the reviews were in perfect english and of similar size and submitted at regular intervals.</p><h4>Results</h4><p>Of the original eleven results only four remained: one from the official AWS website; Whizlabs, who I have used before; and two Kindle books on the Amazon website. So my conclusion is to forget about free questions unless it is from a reputable site. Otherwise you will be dealing with some very scuzzy people.</p><p>Oxford American-English dictionary: scuzzy: <a href="https://www.oxfordlearnersdictionaries.com/definition/american_english/scuzzy">https://www.oxfordlearnersdictionaries.com/definition/american_english/scuzzy</a></p><h3>Free questions from AWS</h3><p>AWS docs: <a href="https://aws.amazon.com/certification/certified-machine-learning-specialty/">https://aws.amazon.com/certification/certified-machine-learning-specialty/</a></p><p>AWS PDF: <a href="https://d1.awsstatic.com/training-and-certification/docs-ml/AWS-Certified-Machine-Learning-Specialty_Sample-Questions.pdf">https://d1.awsstatic.com/training-and-certification/docs-ml/AWS-Certified-Machine-Learning-Specialty_Sample-Questions.pdf</a></p><p>AWS provides a pdf with 10 sample questions. The questions have links to supporting information. These links are to the official AWS website.</p><h3>Free questions from Whizlabs</h3><p>Whizlabs provide 15 questions as a sample of their tests that you can purchase. I have purchased from Whizlabs before and found their products to be of good quality and value for money.</p><p>Whizlabs review: <a href="https://www.mlexam.com/whizlabs-practice-tests/">https://www.mlexam.com/whizlabs-practice-tests/</a></p><h3>Free questions from mlexam.com</h3><p>This website contains 190 free quiz type questions. These cover the content of the exam, but are not in the AWS exam style. Each Study Guide starts with quiz style questions.</p><h4>Sample free quiz</h4><p>This sample quiz has 5 questions randomly chosen from the 190 questions in the Study Guide’s quizzes.</p><p><a href="https://www.mlexam.com/free-questions-aws-machine-learning-exam/#sample-free-quiz">Free questions for the AWS Machine Learning exam - ML exam practice questions</a></p><p>If you liked these quiz style questions why not try some AWS exam style questions: <a href="https://www.mlexam.com/aws-exam-questions/">AWS exam style questions</a></p><h3>AWS practice exam</h3><p>In addition to the standard exam, AWS also provides a practice exam of 40 questions. This exam is taken online and lasts one hour. The cost of the exam is $40. To take a practice exam log into your training account: <a href="https://www.aws.training/certification?src=exam-prep">https://www.aws.training/certification?src=exam-prep</a></p><h3>A Cloud Guru</h3><p>A Cloud Guru provides a practice exam of 65 questions as part of their AWS Machine Learning course. I have used A Cloud Guru to pass two AWS certifications and I can vouch for the quality, relevance and usefulness of their material. They have started to reverse a price hike when they purchased Linux Academy. However this is still a premium service with a price to match.</p><p>A Cloud Guru website: <a href="https://acloudguru.com/">https://acloudguru.com/</a></p><h3>Whizlabs</h3><p>Whizlabs sells 271 questions split into three exams of 65 questions each and a further 76 questions focusing on specific subject areas. Their pack also has four labs. The questions are delivered through an exam app on their website. The answers have detailed explanations and links to specific pages on the official AWS website. I have purchased these practice questions two others in the past and this was an important asset to passing the exam. Whizlabs exam questions are good quality and good value for money.</p><p>Whizlabs review: <a href="https://www.mlexam.com/whizlabs-practice-tests/">https://www.mlexam.com/whizlabs-practice-tests/</a></p><h3>Books on Amazon</h3><h4>AWS Certified Machine Learning Specialty Practice Tests 2020 MLS-C01: Prepare for and pass the current AWS Machine Learning Specialty Exam</h4><p>Contains 130 questions in two exams. It is free if you have a kindle unlimited subscription. 45 questions can be seen by using the look inside feature, although with no answers visible.</p><h3>Conclusion</h3><p>Whilst there are a few opportunities to obtain free exam style questions, I suggest you accept you have to buy from a reputable vendor. The exam will cost $300 so you may as well spend a bit more to get access to this valuable resource.</p><h4>Credits</h4><p>Photo by <a href="https://unsplash.com/@evan__bray?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Evan Dennis</a> on <a href="https://unsplash.com/s/photos/questions?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></p><p><em>Originally published at </em><a href="https://www.mlexam.com/free-questions-aws-machine-learning-exam/"><em>https://www.mlexam.com</em></a><em> on January 18, 2023.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4fdb9e468908" width="1" height="1" alt=""><hr><p><a href="https://medium.com/aws-machine-learning-for-busy-people/free-questions-for-the-aws-machine-learning-exam-4fdb9e468908">Free questions for the AWS Machine Learning exam</a> was originally published in <a href="https://medium.com/aws-machine-learning-for-busy-people">AWS Machine Learning for busy people</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SageMaker BlazingText Algorithm]]></title>
            <link>https://medium.com/aws-machine-learning-for-busy-people/sagemaker-blazingtext-algorithm-26eda7e450cf?source=rss----3e32b8da2ff4---4</link>
            <guid isPermaLink="false">https://medium.com/p/26eda7e450cf</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[sagemaker]]></category>
            <category><![CDATA[mls-c01]]></category>
            <category><![CDATA[aws-certified]]></category>
            <dc:creator><![CDATA[Michael Stainsbury]]></dc:creator>
            <pubDate>Sat, 08 Jul 2023 11:34:00 GMT</pubDate>
            <atom:updated>2023-07-09T09:00:58.013Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jJ94e7yfYB5EbMJcjAf2hQ.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@gasparuhas?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Gaspar Uhas</a> on <a href="https://unsplash.com/@gasparuhas?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>BlazingText is the name AWS has given it’s SageMaker built-in algorithm that can identify relationships between words in text documents. These relationships, which are also called embeddings, are expressed as vectors. The semantic relationship between words is preserved by the vectors which cluster words with similar semantics together. This conversion of words to meaningful numeric vectors is very useful for Natural Language Processing which requires input data in vector format. This is why BlazingText is used as a precursor for Natural Language Processing.</p><p>Word2Vec is used to pre-process documents containing text to be used by other systems. For example: sentiment analysis; machine language from one language to another. Word2Vec generates a numerical representation of words called embedding. This captures the relationships between words so king, queen and president would be closely related. These relationships are used by Natural Language processing systems. BlazingText is an implementation of the Word2Vec algorithm. Word2Vec was published by Google in 2013 and is compatible with Facebook’s FastText.</p><p>Text Classification is used to classify documents, search engines and for document ranking. Text Classification uses embeddings generated by Word2Vec.</p><p>This article contains revision notes for the AWS certified exam MLS-C01, <em>Machine Learning — Specialty</em>.</p><h3>What does the BlazingText algorithm do</h3><p>BlazingText is used for Textual analysis and text classification problems. BlazingText is the only SageMaker built in algorithm to have both <a href="https://www.mlexam.com/unsupervised-learning/">Unsupervised</a> and <a href="https://www.mlexam.com/supervised-learning-for-machine-learning/">Supervised</a> learning modes. Word2Vec is Unsupervised and Text Classification is Supervised learning.</p><ol><li>Word2Vec — Unsupervised learning</li><li>Text Classifier — Supervised learning</li></ol><p>Usually for Text Classification you would pre-process the data by passing it through a Word2Vec algorithm and then a Text Classifier. The BlazingText algorithm implements the Word2Vec and Text Classifier as a single process.</p><h3>How is BlazingText implemented</h3><p>BlazingText is a SageMaker built-in algorithm and so can be trained via SageMaker Jupyter Notebooks and deployed on SageMaker endpoints. Blazing Test processes text data. The input data is presented in a single file with one sentence per line.</p><h3>What are the training data formats for BlazingText</h3><p>There are two input file formats:</p><ol><li>File Mode</li><li>Augmented Manifest Text (AMT) format</li></ol><p>The data in File Mode is text with space separated words and one sentence per line. Each line begins with a label like this: __label__1. The data in Augmented Manifest Text format is in JSON (json lines) format. Each line can contain a single sentence or be split up into phrases by commas as a JSON array. Here are some examples:</p><p>A single line in File Mode:</p><pre>__label__1 Our aim is to increase the year-round consumption of berries in the UK, working closely with British growers during the spring and summer months, and collaborating with UK importers and overseas exporters during winter and early spring.</pre><p>A single JSON line in Augmented Manifest Text format:</p><pre>{&quot;source&quot;:&quot;Our aim is to increase the year-round consumption of berries in the UK, working closely with British growers during the spring and summer months, and collaborating with UK importers and overseas exporters during winter and early spring&quot;,&quot;label&quot;:1}</pre><p>A single JSON array containing phrases in Augmented Manifest Text format:</p><pre>{&quot;source&quot;:&quot;Our aim is to increase the year-round consumption of berries in the UK, working closely with British growers during the spring and summer months, and collaborating with UK importers and overseas exporters during winter and early spring&quot;,&quot;label&quot;:[1,3]}</pre><h3>Model artifacts and inference</h3><p>Blazing Text uses different artifacts depending on it’s processing mode. This table summarises the file names and formats.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/c8d7f5cdf64aff67c9a535dc70291e4b/href">https://medium.com/media/c8d7f5cdf64aff67c9a535dc70291e4b/href</a></iframe><h3>Processing environment</h3><p>BlazingText can be run on a single CPU or GPU instance, or multiple CPU instances. The choice depends on the type of processing being performed. Word2Vec has three processing methods:</p><ol><li>Skip-gram</li><li>Continuous Bag Of Words (CBOW)</li><li>Batch Skip-gram</li></ol><p>These modes are the complete opposites to each other. In skip-gram mode you supply a word and the model returns the context of the word. With CBOW you provide the context and a predicted word is returned.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/766b7e88df4c4f5f72c05f31eaccb052/href">https://medium.com/media/766b7e88df4c4f5f72c05f31eaccb052/href</a></iframe><p>From this table you can see that all processing methods can be performed on a single CPU instance. Only Word2Vec using batch skip-gram method can run on multiple CPUs and this method cannot utilise GPUs.</p><h3>What are BlazingText’s strengths and weaknesses</h3><p>The strength of BlazingText is high performance. BlazingText is more than 20x faster than other popular alternatives such as Facebook’s FastText. This enables inferences to be done in real time for online transactions rather than batch processing. The main weakness of BlazingText is handling words that were not presented during training. These are called Out Of Vocabulary (OOV) words. Typically such words will be marked as Unknown. There are other ways to perform Word2Vec processing, but they do not have the high performance of BlazingText.</p><h3>What is the Use Case for BlazingText</h3><p>BlazingText can only ingest words, so the input data must be text. Word2Vec is required to convert data to vectors for Natural Language Processing.</p><p>Word2Vec:</p><ul><li>Sentiment analysis</li><li>Named entity recognition</li><li>Machine translation</li></ul><p>Text classification:</p><ul><li>Web searches</li><li>Information retrieval</li><li>Ranking</li><li>Document classification</li></ul><h4>Video: AWS re:Invent 2019: Natural language modeling with Amazon SageMaker BlazingText algorithm (AIM375-P)</h4><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FE04JXdnPN0w%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DE04JXdnPN0w&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FE04JXdnPN0w%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/f10008d47faa29933c6d1244df803862/href">https://medium.com/media/f10008d47faa29933c6d1244df803862/href</a></iframe><p>This is a 50.36 minutes video from AWS by Denis Batalov (Linkedin profile: <a href="https://www.linkedin.com/in/denis-v-batalov-59a3111/">https://www.linkedin.com/in/denis-v-batalov-59a3111/</a>). The presentation can be split into four parts as shown in the timestamps below. I suggest you skip the first two parts and start with the overview of SageMaker BlazingText at 17.13 minutes. This is the link to the Jupyter Notebook used in the demo (part 4):</p><ul><li>SageMaker notebook on Github: <a href="https://github.com/dbatalov/wikipedia-embedding">https://github.com/dbatalov/wikipedia-embedding</a></li><li>0 — <strong>Introduction</strong></li><li>2.17 — Word embedding</li><li>2.56 — Word representations</li><li>3.43 — One hot encoding</li><li>4.37 — Intuition, given a sentence, try to maximise the probability of predicting the context of words.</li><li>6.20 — Word2Vec algorithm</li><li>8.20 — t-SNE diagram</li><li>9.23 <strong>— Overview of Amazon SageMaker</strong></li><li>12.20 — Build, train and deploy ML Models</li><li>13.16 — Built-in algorithms</li><li>14.10 — Deep learning frameworks</li><li>15.17 — Automatic Model Tuning</li><li>16.27 — Amazon SageMaker Neo</li><li>17.13 — <strong>Overview of SageMaker BlazingText</strong></li><li>18.28 — BlazingText highlights</li><li>18.45 — Optimization on CPU negative samples sharing</li><li>19.40 — Through characteristics</li><li>20.35 — BlazingText benchmarking</li><li>23.00 <strong>— Demo — Georgian Wikipedia</strong></li></ul><h3>Selected articles with examples of BlazingText being used</h3><p>This article, by Evan Harris, describes the usefulness of having a website search feature tuned to the specific vocabulary used on the website. The example Evan uses is for a search for a specific grape variety which returns a list of wines that use that variety.</p><ul><li><a href="https://medium.com/building-ibotta/heating-up-word2vec-blazingtext-for-real-time-search-c2121bd1396">https://medium.com/building-ibotta/heating-up-word2vec-blazingtext-for-real-time-search-c2121bd1396</a></li></ul><p>This article has a good worked example of BlazingText being used:</p><ul><li><a href="https://t-redactyl.io/blog/2020/09/training-and-evaluating-a-word2vec-model-using-blazingtext-in-sagemaker.html">https://t-redactyl.io/blog/2020/09/training-and-evaluating-a-word2vec-model-using-blazingtext-in-sagemaker.html</a></li></ul><p>This article is a worked example of using BlazingText in Word2Vec mode: <em>Training Word Embeddings On AWS SageMaker Using BlazingText</em> by Roald Schuring.</p><ul><li><a href="https://towardsdatascience.com/training-word-embeddings-on-aws-sagemaker-using-blazingtext-93d0a0838212">https://towardsdatascience.com/training-word-embeddings-on-aws-sagemaker-using-blazingtext-93d0a0838212</a></li></ul><p>This example, from AWS, uses a method to enable BlazingText to generate vectors for out-of-vocabulary (OOV) words.</p><ul><li><a href="https://sagemaker-examples.readthedocs.io/en/latest/introduction_to_amazon_algorithms/blazingtext_word2vec_subwords_text8/blazingtext_word2vec_subwords_text8.html">https://sagemaker-examples.readthedocs.io/en/latest/introduction_to_amazon_algorithms/blazingtext_word2vec_subwords_text8/blazingtext_word2vec_subwords_text8.html</a></li></ul><p>This is an example SageMake Notebook on Github which uses a dataset derived from Wikipedia.</p><ul><li><a href="https://github.com/aws/amazon-sagemaker-examples/blob/master/introduction_to_amazon_algorithms/blazingtext_text_classification_dbpedia/blazingtext_text_classification_dbpedia.ipynb">https://github.com/aws/amazon-sagemaker-examples/blob/master/introduction_to_amazon_algorithms/blazingtext_text_classification_dbpedia/blazingtext_text_classification_dbpedia.ipynb</a></li></ul><h4>Video: Amazon SageMaker’s Built-in Algorithm Webinar Series: Blazing Text</h4><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FG2tX0YpNHfc%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DG2tX0YpNHfc&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FG2tX0YpNHfc%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/7217c1f5a9877ae3fbda0d98ad1af71a/href">https://medium.com/media/7217c1f5a9877ae3fbda0d98ad1af71a/href</a></iframe><p>This is a 1.14.36 minutes video from AWS by Pratap Ramamurthy (Linked in profile: <a href="https://www.linkedin.com/in/pratapramamurthy/">https://www.linkedin.com/in/pratapramamurthy/</a>). This is a very long video so use the timestamps below to select the part you wish to see.</p><ul><li>0 — Introduction</li><li>2.19 — What are Amazon algorithms</li><li>3.08 — BlazingText algorithms</li><li>3.17 — BlazingText use case</li><li>4.16 — Typical deep learning task on Text</li><li>5.36 — Integer encoding</li><li>9.20 — One hot encoding</li><li>14.00 — Requirements for word vectors</li><li>16.32 — Word2Vec mechanism</li><li>16.42 — Word2Vec setup</li><li>18.07 — Skip-gram preprocessing</li><li>20.30 — Neural network setup</li><li>25.38 — BlazingText word embedding</li><li>27.35 — Word vectors used for further ML training</li><li>28.20 — Intuition</li><li>28.25 — Random or is there a pattern? (t-SNE plot)</li><li>31.14 — Distance between related words</li><li>32.26 — How did the magic work?</li><li>35.08 — OOV handling using BlazingText</li><li>39.38 — Subword detection</li><li>41.43 — Text classification with BlazingText</li><li>42.18 — Typical NLP pipeline</li><li>44.25 — Parameters</li><li>47.43 — Demo</li><li>100.11 — Questions</li></ul><h3>Conclusion</h3><p>BlazingText is a high performance algorithm for analyzing text. The two modes of processing producing either numeric vectors for Natural Language Processing via the Word2Vec algorithm or Text Classifications that can infer words from context or context from words.</p><h3>Resources</h3><p>These revision notes support subdomain 3.2 <em>Select the appropriate model(s) for a given machine learning problem</em> of the AWS certification exam: <em>AWS Machine Learning — Speciality</em> (MLS-C01).</p><blockquote><em>3.2 Select the appropriate model(s) for a given machine learning problem.<br>Xgboost, logistic regression, K-means, linear regression, decision trees, random forests, RNN, CNN, Ensemble, Transfer learning. Express intuition behind models</em></blockquote><p><em>AWS Certified Machine Learning — Specialty, (MLS-C01) Exam Guide</em></p><ul><li>AWS Certified Machine Learning exam guide: <a href="https://www.mlexam.com/aws-machine-learning-exam-guide/">www.mlexam.com/aws-machine-learning-exam-guide</a></li><li>3 Modeling: <a href="https://www.mlexam.com/home/domain-3-modeling/">www.mlexam.com/home/domain-3-modeling/</a></li><li>3.2 Text processing algorithms: <a href="https://www.mlexam.com/sagemaker-text-processing-algorithms/">www.mlexam.com/sagemaker-text-processing-algorithms</a></li><li>Questions for SageMaker built-in algorithms and their uses: <a href="https://www.mlexam.com/35-q-a-for-sagemaker-built-in-algorithms/">www.mlexam.com/35-q-a-for-sagemaker-built-in-algorithms</a></li><li>Free Practice exam: <a href="https://www.mlexam.com/aws-machine-learning-practice-exam/">www.mlexam.com/aws-machine-learning-practice-exam</a></li></ul><p>Overview</p><ul><li>AWS docs: <a href="https://docs.aws.amazon.com/sagemaker/latest/dg/blazingtext.html">https://docs.aws.amazon.com/sagemaker/latest/dg/blazingtext.html</a></li><li>Wikipedia Word2vec: <a href="https://en.wikipedia.org/wiki/Word2vec">https://en.wikipedia.org/wiki/Word2vec</a></li><li>Google original papers from 2013: <a href="https://arxiv.org/abs/1301.3781">https://arxiv.org/abs/1301.3781</a></li><li>Google original papers from 2013: <a href="https://arxiv.org/abs/1310.4546">https://arxiv.org/abs/1310.4546</a></li></ul><p>Training data format resources</p><ul><li>Augmented Manifest Text (AMT) format: <a href="https://docs.aws.amazon.com/sagemaker/latest/dg/augmented-manifest.html">https://docs.aws.amazon.com/sagemaker/latest/dg/augmented-manifest.html</a></li><li>Json lines format: <a href="http://jsonlines.org/">http://jsonlines.org/</a></li><li>Text examples from <a href="https://www.britishsummerfruits.co.uk/about">https://www.britishsummerfruits.co.uk/about</a></li></ul><p>Processing environment</p><ul><li><a href="https://aws.amazon.com/blogs/machine-learning/enhanced-text-classification-and-word-vectors-using-amazon-sagemaker-blazingtext">https://aws.amazon.com/blogs/machine-learning/enhanced-text-classification-and-word-vectors-using-amazon-sagemaker-blazingtext</a></li></ul><h4>Credits</h4><p>Burning book photo by<a href="https://unsplash.com/@gasparuhas?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Gaspar Uhas</a> on<a href="https://unsplash.com/s/photos/writing-on-fire?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Unsplash</a></p><p><em>Originally published at </em><a href="https://www.mlexam.com/blazingtext-algorithm/"><em>https://www.mlexam.com</em></a><em> on March 2, 2021.</em></p><p><em>First published in Medium on July 6, 2023.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=26eda7e450cf" width="1" height="1" alt=""><hr><p><a href="https://medium.com/aws-machine-learning-for-busy-people/sagemaker-blazingtext-algorithm-26eda7e450cf">SageMaker BlazingText Algorithm</a> was originally published in <a href="https://medium.com/aws-machine-learning-for-busy-people">AWS Machine Learning for busy people</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>