<?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[Stories by Keerthika Sathish on Medium]]></title>
        <description><![CDATA[Stories by Keerthika Sathish on Medium]]></description>
        <link>https://medium.com/@keerthikasathish?source=rss-67633b604ea4------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Stories by Keerthika Sathish on Medium</title>
            <link>https://medium.com/@keerthikasathish?source=rss-67633b604ea4------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 11:07:12 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@keerthikasathish/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[16,700 INR Scholarship from AWS]]></title>
            <link>https://medium.com/@keerthikasathish/my-journey-of-receiving-a-scholarship-from-aws-valued-at-16-700-inr-4f3d3db375fd?source=rss-67633b604ea4------2</link>
            <guid isPermaLink="false">https://medium.com/p/4f3d3db375fd</guid>
            <dc:creator><![CDATA[Keerthika Sathish]]></dc:creator>
            <pubDate>Mon, 26 Feb 2024 15:15:29 GMT</pubDate>
            <atom:updated>2025-01-23T13:59:07.970Z</atom:updated>
            <content:encoded><![CDATA[<p>I was one among 1000 applicants from around the world chosen for the program, and for that, I would like to thank my friend, Madhu Vadhani, who informed me about the AWS x Udacity NanoDegree scholarship program.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/0*ejqfnH0RHCZQ2VYW" /><figcaption>from Edu Tech</figcaption></figure><p>The acceptance process for the course comprises two main components:</p><ol><li>creating a simple project</li></ol><p>2. Respond to a set of questions.</p><p>Initially, fundamental concepts are imparted, and participants are tasked with modifying a given project. The primary objective is to optimize the project to minimize overall completion time. Subsequently, participants are queried on various aspects, focusing on the ‘what’ and ‘why’ of the learned concepts.</p><p>I completed the program in March 2023, during which I was asked to reduce the completion time for a race track task. This involved implementing reinforcement learning principles, enabling the model to learn from its errors.</p><h3>About the program</h3><p>The program delved into fundamental concepts essential for machine learning, providing insights into both simple and complex structures. It explains the significance of utilizing linear algebra, calculus, and various other mathematical tools.</p><p>Throughout each module, projects were assigned, culminating in a final project after the course.</p><h3>Instructor Support</h3><ul><li>Udacity provides access to instructors and mentors who address queries and offer guidance.</li><li>I extend my gratitude to my mentor, Sir Ali Zawahry, for his endless support and guidance.</li></ul><p>Similar scholarship programs can be found <a href="https://aws.amazon.com/machine-learning/scholarship/#">here</a>.</p><p>By the end of the program, I had acquired a comprehensive understanding of the fundamental principles underlying machine learning and artificial intelligence.</p><p><em>If you liked this article, be sure to click ❤ below to recommend it and if you have any questions, leave a comment.</em></p><p><em>I hope you like this blog and thanks for your time in reading it 😄. You can follow me on </em><a href="https://www.linkedin.com/in/keerthika-sat/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4f3d3db375fd" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Create, Play, Imagine: Generative]]></title>
            <link>https://medium.com/@keerthikasathish/create-play-imagine-generative-050769936959?source=rss-67633b604ea4------2</link>
            <guid isPermaLink="false">https://medium.com/p/050769936959</guid>
            <dc:creator><![CDATA[Keerthika Sathish]]></dc:creator>
            <pubDate>Sat, 27 Jan 2024 05:56:14 GMT</pubDate>
            <atom:updated>2024-01-27T05:56:14.656Z</atom:updated>
            <content:encoded><![CDATA[<h3>Create, Play, Imagine: Generative AI</h3><p>During my childhood, I used to draw all sorts of random stuff, and even now, I love it. As I grew older, I started using Microsoft Paint, then moved to Paint 3D and other software. I love to explore the tools. That’s how I stumbled upon Generative AI.</p><p>Generative AI is nothing but a tool where users can quickly create new content based on the inputs. Initially, it seems to be boring; however, when engaging with the software, curiosity about its function arises.</p><p>Here’s an AI image generated by <a href="https://www.bing.com/images/create">Image Creator from Microsoft Designer </a>by using the text description “a woman in a music festival facing the crowd, with a glittery dress twirling around and bright dress, semi-wavy hair, digital art”.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/293/1*lgRx-bgNK0a-fcE6YIviPA.png" /></figure><p>There are various techniques for generating an image. The most popular and widely technique is <strong>Generative Adversarial Network</strong> aka GAN. Let’s look into its functions and architecture.</p><h4><strong>Generative Adversarial Network (GAN)</strong></h4><p>GAN consists of two main components: the generator and the discriminator.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5yp_VfSL762ElneXSf8qEw.png" /></figure><p><strong>1. Generator:</strong></p><ul><li>It takes random input data and is composed of CNN (Convolutional Neural Network) or RNN (Recurrent Neural Network)</li><li>It creates numerous fake images.</li></ul><p><strong>2. Discriminator:</strong></p><ul><li>The discriminator evaluates the realness of the image, determining whether they are real or fake.</li><li>It is also composed of deep neural networks like CNN, RNN or any other discriminator models.</li><li>Receives both real and fake image samples and fake samples generated by the generator.</li></ul><p>Then the model provides discriminator loss measures how well it distinguishes between real and fake data, while the loss of the generator indicates how effectively it generates data that fools the discriminator. Then the loss value is backpropagated to its respective ones and the model stops functioning effectively when the generator produces data indistinguishable from real data, causing the discriminator to struggle to differentiate between real and fake samples.</p><p>It indicates that the generator and discriminator have reached a level of performance where further improvements are minimal.</p><p>The model has to be trained with the highest number of epochs to create a fake image that is as real as a real image. The generator aims to fool the discriminator, by generating a realistic image. The epoch size depends upon various conditions like the complexity of the dataset, the architecture of the model, and the optimization technique.</p><p><em>If you liked this article, be sure to click ❤ below to recommend it and if you have any questions, leave a comment.</em></p><p><em>I hope you like this blog and thanks for your time in reading it 😄. You can follow me on </em><a href="https://www.linkedin.com/in/keerthika-sat/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=050769936959" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What is Clustering?]]></title>
            <link>https://medium.com/@keerthikasathish/what-is-clustering-b1bdd8a19336?source=rss-67633b604ea4------2</link>
            <guid isPermaLink="false">https://medium.com/p/b1bdd8a19336</guid>
            <dc:creator><![CDATA[Keerthika Sathish]]></dc:creator>
            <pubDate>Sun, 26 Nov 2023 08:39:25 GMT</pubDate>
            <atom:updated>2023-11-26T08:39:25.612Z</atom:updated>
            <content:encoded><![CDATA[<p>In general, clustering means the grouping of similar objects. Likewise, it means the same in machine learning, it is the grouping of similar unlabeled data. It is an unsupervised machine-learning technique.</p><h4>Types of Clustering</h4><ol><li>Density-based clustering</li><li>Hierarchy clustering</li><li>Partitioning clustering</li></ol><h4>Density-based clustering</h4><p>It is where the data points are clustered based on their density or weight. The data points with higher density will be grouped into one while the others are grouped separately.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/220/0*qqffmdS812XzelvR.png" /><figcaption>Wikipedia</figcaption></figure><h4>Hierarchy clustering</h4><p>The dataset is divided into clusters to create a tree-like structure, which is also called a dendrogram. The tree is grouped based on the distance between the data points.</p><p>The knee method is used to break the cluster and identify the k-value</p><p>k =&gt; number of clusters</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/0*Uu2B1ljKovVIaHpX" /><figcaption>Spot Intelligence</figcaption></figure><h4>Partitioning clustering</h4><p>It is where the data points are divided into partitions. It uses the centroid method to divide the data. The centroid is calculated using Euclidean distance. The k-means clustering algorithm is the popular method for partitioning clustering.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/239/0*R89Pfrg7yZfGpWdo" /><figcaption>Google for developers</figcaption></figure><h4>Partitioning vs Hierarchical Clustering</h4><p>a =&gt; partitioning clustering</p><p>b =&gt; heirarchicala clustering</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/591/0*yijEudGXPmkZpokz.png" /><figcaption>InTech Open</figcaption></figure><p><em>If you liked this article, be sure to click ❤ below to recommend it and if you have any questions, leave a comment.</em></p><p><em>I hope you like this blog and thanks for your time in reading it 😄. You can follow me on </em><a href="https://www.linkedin.com/in/keerthika-sat/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b1bdd8a19336" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Mastering NLP for AI900]]></title>
            <link>https://medium.com/@keerthikasathish/mastering-nlp-for-ai900-3d8f8670431d?source=rss-67633b604ea4------2</link>
            <guid isPermaLink="false">https://medium.com/p/3d8f8670431d</guid>
            <dc:creator><![CDATA[Keerthika Sathish]]></dc:creator>
            <pubDate>Sat, 23 Sep 2023 14:24:29 GMT</pubDate>
            <atom:updated>2023-09-23T14:24:29.477Z</atom:updated>
            <content:encoded><![CDATA[<p>I’ll share my preparation and tips for mastering Natural Language Processing (NLP) for AI900.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*GaHrEHbDjHOQlo1F.jpg" /><figcaption>Source: Analytics Vidhya</figcaption></figure><p>NLP is the ability of computers to understand human language(spoken and written).</p><p><strong>Sentiment Analysis — </strong>the language model predicts whether a sentence gives a positive, negative or neutral impact.</p><p><strong>Key phrase extraction — </strong>extracts the input words from the document.</p><p><strong>Entity —</strong> An entity is essentially an item of a particular type or a category. For example</p><p>Bill Gates, John are Person. Where person is entity</p><p>Bangalore, Chennai — Location(entity)</p><p><strong>Speech to text and text to speech API — </strong>The language model helps to convert human speech to text and text to speech respectively. Google Assistant is an example of speech to text API.</p><p>The language model can also generate spoken responses from the text.</p><p><strong>Speech translation — </strong>Where the language model can understand multiple languages and respond in your preferred language. It can understand in one language and respond in another language. Google translator can translate text from one language to another.</p><p><strong>Intent —</strong> It is helps to find out underlying purpose or goal that a person expresses when they communicate. For example</p><p>Switch on the fan</p><p>Intent -&gt; on</p><p>Entity -&gt; fan</p><h4>Entity types:</h4><ul><li>RegEx: Entities that are defined as a <em>regular expression</em> that describes a pattern — for example, you might define a pattern like [0–9]{3}-[0–9]{3}-[0–9]{4} for telephone numbers of the form <em>555–123–4567</em>.</li><li>Machine-Learned: Entities that are learned by your model during training from context in the sample utterances you provide.</li><li>List: Entities that are defined as a hierarchy of lists and sublists. For example, a device list might include sublists for light and fan. For each list entry, you can specify synonyms, such as lamp for light.</li><li>Pattern.any: Entities that are used with <em>patterns</em> to define complex entities that may be hard to extract from sample utterances.</li></ul><p>Any NLP model has to be fed with data and trained to predict accurate output.</p><p>These are the fundamental NLP concepts that you need to learn to excel in AI900.</p><p><em>If you liked this article, be sure to click ❤ below to recommend it and if you have any questions, leave a comment.</em></p><p><em>I hope you like this blog and thanks for your time in reading it 😄. You can follow me on </em><a href="https://www.linkedin.com/in/keerthika-sat/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3d8f8670431d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Microsoft Learn Student Ambassador journey]]></title>
            <link>https://medium.com/@keerthikasathish/my-journey-of-becoming-microsoft-learn-student-ambassador-mlsa-ca6e04d87545?source=rss-67633b604ea4------2</link>
            <guid isPermaLink="false">https://medium.com/p/ca6e04d87545</guid>
            <category><![CDATA[aml]]></category>
            <category><![CDATA[swag]]></category>
            <category><![CDATA[microsoft]]></category>
            <dc:creator><![CDATA[Keerthika Sathish]]></dc:creator>
            <pubDate>Thu, 07 Sep 2023 08:27:12 GMT</pubDate>
            <atom:updated>2023-09-07T15:40:22.437Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Do you find it difficult to become Microsoft Learn Student Ambassador (MLSA), don’t worry you have got me.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/268/1*vspViUxabUE8Cx7vHWHY5Q.png" /></figure><p>Regardless of your background, whether you’re a student enrolled full-time in an accredited academic institution — a fresher or a final year — all that matters is your enthusiasm and passion towards technology.</p><p>If you’re interested in learning the steps to join the program, you can find the enrollment procedure on the <a href="https://mvp.microsoft.com/studentambassadors">website</a> or explore various blogs. My goal is to share my own experiences with the community and assist other students who may have faced challenges in their previous attempts to join the program.</p><p>You can enroll in <a href="https://mvp.microsoft.com/studentambassadors">Microsoft Learn Student Ambassador</a>. By enrolling you can -</p><ol><li>Upskill yourself with Microsoft courses, get mentored by Microsoft professionals, earn badges and certificates.</li><li>Host events, workshops, and hackathons at your college teach people how to code, and become a LEADER.</li><li>Network with the global community.</li></ol><h4>My journey towards Microsoft Learn Student Ambassador</h4><p>During my first year of college, I applied for the Microsoft Learn Student Ambassador (MLSA) program unfortunately, I was not selected. However, I remained confident and decided to give it another shot. This time, I had the support of my sister, <a href="https://www.linkedin.com/in/keerthanasathish/">Keerthana Sathish</a>, who had already been selected for the program. Thanks to her referral, I was able to successfully join the MLSA program.</p><p>The milestones are <strong>New, Alpha, Beta, Gold </strong>student ambassadors</p><p>I was landed into “New” student ambassador and got promoted to Alpha and then Beta student ambassador by following the steps mentioned for upgradation.</p><p>As you advance to next level you’ll unlock new benefits. Once joining you will also receive a student ambassador mail id.</p><p>Well, when you advance to the beta level you get an amazing swag kit as your appreciation and a virtual certificate with you in your MLSA profile. Make sure to download the current level certificate before advancing to the next level because they will replace your current level certificate with your next level certificate.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/727/1*6nCtUkS7n7V8gd1A_nhBZg.png" /></figure><p>For being promoted to Gold you need to be in the community for 1 year. You can host your own sessions or attends others session</p><p>One of the most rewarding aspects of being part of this community is the networking opportunities it offers. You can seek advice, learn about different opportunities, and participate in exclusive events tailored for MLSA students.</p><p>The community also hosts various tech sessions and meetings. Additionally, you’ll be encouraged to participate in a Quarterly Project where you can choose from options like writing a blog, creating a prototype, or organizing an event. This project allows you to collaborate with fellow student ambassadors and further enrich your experience in the program.</p><p>Furthermore, you can take the initiative to establish a Microsoft Student Community at your college. This initiative will enable you to connect with students on your campus.</p><h4>Meetups</h4><p>You’ll also get the chance to attend offline meetups, providing a valuable opportunity to connect with fellow student ambassadors. I had the privilege of attending a meetup in Bangalore where students from diverse regions came together. Meeting them in person was truly exciting and inspiring. These meetups are typically organized by Gold Student Ambassadors, and they often feature tech professionals who share insights into various technologies. At the end, ambassadors were given swag kits.</p><p>Hope this blog will encourage you further.</p><p><em>If you liked this article, be sure to click ❤ below to recommend it and if you have any questions, leave a comment.</em></p><p><em>I hope you like this blog and thanks for your time in reading it 😄. You can follow me on </em><a href="https://www.linkedin.com/in/keerthika-sat/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ca6e04d87545" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A Winning Approach to Mastering AI-900]]></title>
            <link>https://medium.com/@keerthikasathish/a-winning-approach-to-mastering-ai-900-796a90c26ba5?source=rss-67633b604ea4------2</link>
            <guid isPermaLink="false">https://medium.com/p/796a90c26ba5</guid>
            <dc:creator><![CDATA[Keerthika Sathish]]></dc:creator>
            <pubDate>Fri, 18 Aug 2023 12:32:20 GMT</pubDate>
            <atom:updated>2023-09-23T14:26:04.970Z</atom:updated>
            <content:encoded><![CDATA[<h3>Strategies to Ace AI-900</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/847/1*nX67zICuA4cmDqY6bK7exQ.png" /></figure><p>If you want to build your resume and know the basics of AI&amp;ML, this course is one among them.</p><h4>Why AI-900?</h4><p>AI-900 exam is an opportunity for you to demonstrate knowledge of machine learning and AI concepts and related Microsoft Azure services. It not only helps to build a strong foundation on the basics but also the ethics to be followed when handling data. Understanding of basic services provided by Azure.</p><h4>Preparation</h4><ol><li><a href="https://learn.microsoft.com/en-us/certifications/azure-ai-fundamentals/">Microsoft Learn Free Learning Path</a>: This is the official website where Microsoft offers courses for professionals. I suggest going through the modules without skipping and finishing the knowledge check and exercises. This will greatly help in answering many exam questions. Also, there’s a student certificate documentation that students can look into for their benefits.</li><li>YouTube: Channels like “<a href="https://www.youtube.com/watch?v=E2644J4jgd8&amp;list=PLhLKc18P9YOAGdtSpB3y7WMkjJbi7dva_&amp;index=1">A Guide to Cloud</a>” makes it easier to understand the topics with simple explanations.</li><li>Previous questions: Once you’ve learned the course, make sure to take a practice test. This will help you become familiar with the concepts and identify areas where you might need more practice.</li></ol><h4>Types of Questions Asked</h4><ol><li>Multiple Choice Question: You need to choose the right answer from a list of options. There could be one correct answer and three wrong ones. Sometimes, there might be more than one correct answer.T</li><li>True or False: You just have to say if a statement is true or false. You can pick “Yes” for true and “No” for false.</li><li>Drag and Drop: This is like sorting things into the right places. You’ll be given some answers, and you need to drag them to match the correct descriptions.</li><li>Hot Areas: Imagine you have a picture with different parts labeled with questions. You have to interact with the picture by clicking on the right spots to answer the questions.</li></ol><h4>Distribution of Syllabus</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/976/1*S-LNSSb_ZyyHrYQAHDbguA.png" /><figcaption>Source: Microsoft AI-900 Certification Exam (as of February 2021)</figcaption></figure><h4>My Experience</h4><p>Questions about ethics were not very common in the exam. It’s better to focus more on topics like Computer Vision, Machine Learning, and the fundamental principles of Machine Learning. The practical exercises are done on the Azure platform, which helps you learn how to use the Azure portal in real-world situations.</p><p>Understand the ideas well and take a practice test to check how well you know them. This will help you build a solid understanding of the concepts. The exam will ask questions about what you’ve learned, so this practice will prepare you.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LMMrpWoa28TFVu8PIMxRYw.jpeg" /></figure><p>Hope this blog will help you to ace the exam.</p><p><em>If you liked this article, be sure to click ❤ below to recommend it and if you have any questions, leave a comment.</em></p><p><em>I hope you like to blog and thanks for your time in reading it 😄. You can follow me on </em><a href="https://www.linkedin.com/in/keerthika-sat/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=796a90c26ba5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Is AutoML a Game Changer?]]></title>
            <link>https://medium.com/@keerthikasathish/is-automl-a-game-changer-169a21f7006a?source=rss-67633b604ea4------2</link>
            <guid isPermaLink="false">https://medium.com/p/169a21f7006a</guid>
            <dc:creator><![CDATA[Keerthika Sathish]]></dc:creator>
            <pubDate>Sat, 05 Aug 2023 07:36:49 GMT</pubDate>
            <atom:updated>2023-08-05T10:09:01.818Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*W7qcGp5ZTHKIXt5RrhJKjA.jpeg" /><figcaption>source: Photo by <a href="https://unsplash.com/@askkell?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Andy Kelly</a> on <a href="https://unsplash.com/photos/0E_vhMVqL9g?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>AutoML stands for Automated Machine Learning. It is the process of automating the end-to-end process in real-world problems by applying Machine Learning. According to the survey of CBInsights AutoML impacts industry/ jobs by 47%. It accelerated the job of Data Science</p><p>In simple terms, AutoML automates the selection and tuning process of the algorithm. This includes from the time of pre-processing of data till the appropriate model selection for the problem. It comprises <em>data processing, feature generation and selection, and model selection and hyperparameters selection</em>. Nowadays tech giants are opting for AutoML by allowing their employees to focus on complex problems, increase accuracy and decrease the amount of time.</p><h4>Architecture of AutoML</h4><p>If you are wondering about the architecture then here it is. It uses a Neural Architecture Search (NAS) set of algorithms applied to neural networks and deep learning. The biggest advantage of this algorithm is it does require minimum human intervention and a limited amount of resources is used to design the architecture.</p><p>In addition to Neural Architecture Search (NAS), there exist alternative <br>methodologies within the field of Automated Machine Learning (AutoML). Some of the approaches are given below</p><ul><li>Bayesian Optimization</li><li>Evolutionary Algorithm</li><li>Gradient-based methods</li><li>Multi-modal methods</li><li>Meta-learning</li></ul><h4>Machine Learning vs Automatic Machine Learning</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/664/1*EJDvb4bnxhD9m9ApgfTivg.png" /></figure><p><em>If you liked this article, be sure to click ❤ below to recommend it and if you have any questions, leave a comment.</em></p><p><em>I hope you like to blog and thanks for your time in reading it 😄. You can follow me on </em><a href="https://www.linkedin.com/in/keerthika-sat/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=169a21f7006a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[TinyML — the future of machine learning]]></title>
            <link>https://medium.com/@keerthikasathish/tinyml-the-future-of-machine-learning-5c0bd6999e58?source=rss-67633b604ea4------2</link>
            <guid isPermaLink="false">https://medium.com/p/5c0bd6999e58</guid>
            <dc:creator><![CDATA[Keerthika Sathish]]></dc:creator>
            <pubDate>Wed, 21 Jun 2023 11:14:06 GMT</pubDate>
            <atom:updated>2023-06-21T11:14:06.106Z</atom:updated>
            <content:encoded><![CDATA[<h3>TinyML — the future of machine learning</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5pzkKDlj_0cuNsIoNOynAw.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@possessedphotography">Possessed Photography</a> on Unsplash</figcaption></figure><p>We are currently living in a world full of data and machines. Data science and machine learning are one of the revolutionizing technology in the world. Day to day tasks like watching reels, checking on the weather, suggestions on social media, all depend on machine learning models and the data that are provided to train the models. Even this blog might be a suggested one.</p><p>Training the models requires high computational costs and time. The rate at which the world is progressing requires the models to be computationally fast enough to handle it. Thus most of the models run on huge data centers with clusters of CPUs and GPUs(even TPUs when required).</p><p>To meet the requirements of today’s world <strong><em>tinyml </em></strong>is introduced.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/290/1*mEh63-iE1LsiuPfDxU_xkQ.png" /></figure><h4>What is TinyML</h4><p>TinyML or tiny machine learning is an emerging field. It focuses on developing and deployment of machine learning models in low-powered and resource constraint devices, such as microcontrollers. These devices are found in IoT (Internet of Things) applications, embedded devices, and other wearables where computational and high-heavy machine learning models are unsuitable.</p><p>It uses limited computational resources (memory and processing power) available on the devices. The goal is to reduce reliance on cloud services and enable on-device processing by improving privacy, security, and response time.</p><h4>Key aspects</h4><ol><li><strong>Model Efficiency:</strong> TinyML models are designed to be efficient with low memory and computational requirements. Model’s complexity and size are reduced by employing optimization techniques like quantization and model compression.</li><li><strong>Training and Inference: </strong>TinyML involves developing techniques for training models with limited data and computational resources. It also focuses on ensuring the prediction of the model within the device constraint.</li><li><strong>Hardware Constraints: </strong>TinyML models need to run on devices with limited resources, such as microcontrollers with low-power processors and limited memory. These constraints necessitate the development of specialized algorithms and techniques to make machine learning feasible on such devices.</li></ol><h4>Conclusion</h4><p>Machine Learning is all about big data which helps the model to give accurate and precise predictions, but big data cannot be used in every model. TinyML will be revolutionizing how we think about machine learning and big data. As there is advancement in technologies we are able to see significant and innovative applications and use cases, unlocking ML’s full potential.</p><p><em>If you liked this article, be sure to click ❤ below to recommend it and if you have any questions, leave a comment.</em></p><p><em>I hope you like to blog and thanks for your time in reading it 😄. You can follow me on </em><a href="https://www.linkedin.com/in/keerthika-sat/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5c0bd6999e58" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Era of Artificial Neurons]]></title>
            <link>https://medium.com/@keerthikasathish/the-era-of-artificial-neurons-d6df78269b65?source=rss-67633b604ea4------2</link>
            <guid isPermaLink="false">https://medium.com/p/d6df78269b65</guid>
            <dc:creator><![CDATA[Keerthika Sathish]]></dc:creator>
            <pubDate>Mon, 05 Jun 2023 10:13:32 GMT</pubDate>
            <atom:updated>2023-10-05T10:14:14.481Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/607/1*UVP0uMIiR0UwW7HzSMKUng.png" /></figure><p>All of you know what neurons is and how neurons help the brain to function. Neurons are the nerve cells that send messages all over the body. It allows you to do everything from breathing to thinking.</p><p>Artificial neurons are units in a system that imitate the behavior of biological neurons found in the human brain. They receive input information, process it, and produce an output based on that information. Clusters of artificial neurons are called Artificial Neural Networks (ANN).</p><p><strong>Biological Neuron</strong></p><p>o understand the functionality of artificial neural networks, first, you need to know about biological neurons. Biological neurons are nothing but brain neurons. The basic elements of human brain neurons are dendrites, axons, and axon terminals.</p><p>Dendrites are branch-like structures extending from the cell body of a neuron. Their function is to receive incoming signals or information from other neurons. They act as the input region of the neuron, collecting electrical impulses and transmitting them to the cell body.</p><p>An axon is a lengthy and slender projection that emanates from the cell body of a neuron. Its primary function is to transmit electrical signals, known as action potentials, away from the cell body to other neurons or target cells. The axon carries these signals over long distances, allowing communication between various parts of the nervous system.</p><p>Axon terminals are small branches at the end of an axon. Their function is to transmit signals from the axon to other neurons or target cells. Axon terminals form connections, called synapses, with dendrites or cell bodies of other neurons, allowing for communication through the release of chemical neurotransmitters.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*N3YkHbD6ZpjrfGHzplBYmQ.png" /></figure><p><strong>Artificial Neural Network</strong></p><p>A neural networks are inspired by the functioning and structure of human brain. It consists of interconnected artificial neurons that work together to process information. Neural networks are designed to learn from data, make predictions, or solve problems by adjusting the strengths of connections between neurons. They are used in various applications like image recognition, language processing, and decision-making tasks. Neural networks are a way for computers to mimic the learning and decision-making capabilities of the human brain.</p><p>The neural network has three main layers — an input layer, a hidden layer, and an output layer.</p><p>The input layer is responsible for receiving the input data from the system and passing it to the hidden layer. Each neuron in input layers corresponds to features or attributes of the input data</p><p>The hidden layer is the intermediate layer between the input layer and the output layer. It performs computations on the input layer and passes it on to the output layer. The hidden layer can have as many layers as possible. A neural network with a small number of hidden layers, or even without any hidden layers, may struggle to capture complex patterns and relationships in the data. Whereas high number of hidden layers causes high computational time, and the network will be prone to overfitting</p><p>Overfitting occurs when a system captures patterns and details from the training data to the extent that it fails to generalize well to new, unseen data. It is like memorizing the answers to a specific set of questions without utterly understanding the underlying concepts.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/422/1*jWzmB_JSytsV0mm70fziOQ.png" /></figure><p><strong>Conclusion</strong></p><p>Neural networks are the future. Their ability to learn by example makes them powerful. They are well suited for real-world systems because of their fast computational time.</p><p>Neural networks contribute a lot to research and have found various applications in various domains such as image recognition, data analysis, neuroscience, data analysis, and much more. They have revolutionized industries by automating tasks, improving efficiency, and providing insights from vast amounts of data.</p><p><em>If you liked this article, be sure to click ❤ below to recommend it and if you have any questions, leave a comment.</em></p><p><em>I hope you like to blog and thanks for your time in reading it 😄. You can follow me on </em><a href="https://www.linkedin.com/in/keerthika-sat/"><em>LinkedIn</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d6df78269b65" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>