<?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 Horn Daneth on Medium]]></title>
        <description><![CDATA[Stories by Horn Daneth on Medium]]></description>
        <link>https://medium.com/@horndaneth?source=rss-1dd1391a900d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*mYw91NuH034fuq61BaMyyg.png</url>
            <title>Stories by Horn Daneth on Medium</title>
            <link>https://medium.com/@horndaneth?source=rss-1dd1391a900d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 08:59:28 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@horndaneth/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[Beautiful Morning of My Hometown]]></title>
            <link>https://medium.com/@horndaneth/beautiful-morning-of-my-hometown-adb288c82153?source=rss-1dd1391a900d------2</link>
            <guid isPermaLink="false">https://medium.com/p/adb288c82153</guid>
            <category><![CDATA[hometown]]></category>
            <dc:creator><![CDATA[Horn Daneth]]></dc:creator>
            <pubDate>Sat, 04 May 2019 15:16:24 GMT</pubDate>
            <atom:updated>2019-06-10T14:10:08.908Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/605/1*GdviVqOQb9dK_CMYepwi0g.png" /></figure><p>In a beautiful morning from the top of the highest mountain, the darkness inevitably vanishes, and the golden light finally comes out from the shining sun on the surface of the verdant mountain. A diminutive house on the east side of the mountain welcomingly smiles at a new day. Thousands of animals living on the mountain loudly sing morning songs and brutally wake the villagers up from a deep slumber. A noisy flock of starving sparrows arise and leave their nest to forage for foods.</p><p>On the north side of the mountain, a skimpy river lazily strolls along a teeny village of the mountain like a python slowly crawling across boulevard at dawn. A big crowd of fishes hastily emerge from deep water and happily reveal themselves to the bright sun. Suddenly, a gust of wind blows and gently turns the surface of the river into choppy layers.</p><p>After a few moments, atop the mountain, visitors leisurely start to come and feed food to small animals. Along the bride, sellers energetically start to set up shops to sell magnificent souvenirs. A quiet bridge between the quiet mountain and noisy village turns into a heavy buzzing, and a gigantic sculpture of a lion on the bridge stands lonely by the gate and sadly watches a convoy of visitor cars.</p><p>At the end of the bridge, a lovely girl runs into a baby sparrow eating a blade of grass and chirps, “Good morning. Where is your mother?”. The baby bird instantly rushes up to the tiny nest and poorly weeps, “chhirp chhirp chhirp.” The little girl curiously follows the poor bird to the nest and dismally looks at the small bird and says, “ Are you hungry?” and gives a bundle of cookies to the baby sparrow. After a minute, the bird’s mother comes to comfort the baby. The baby sparrow starts smiling and dances in the small nest.</p><p>In the blink of an eye, a happy moment of the poor sparrow, however, change into pathetic frightened when the dark clouds massed on the beautiful mountain. Then, heavy thunderstorm sounds with heavy downpours of rain and loud thunder. The small sparrow cries sorrowfully once again. The sparrow’s mother and the poor sparrow quickly seek out shelter to avoid getting wet and cold. Dreadfully, the birds cannot figure out any shelters. The relief is instantaneous when the generous girl runs into the poor birds again. The small girl takes both birds and puts under eaves of the tiny house on the mountain. The baby bird starts feeling better and has a big grin on her face.</p><p>After a while, the sky passes over the sun, and it turns to a perfect blue, not any clouds in sight. The loud sound of animals blares out on the mountain. Once again, the beautiful morning smiles and invites to visitors.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=adb288c82153" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Machine learning interview questions]]></title>
            <link>https://medium.com/@horndaneth/machine-learning-interview-questions-edcc753dd9af?source=rss-1dd1391a900d------2</link>
            <guid isPermaLink="false">https://medium.com/p/edcc753dd9af</guid>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Horn Daneth]]></dc:creator>
            <pubDate>Wed, 14 Nov 2018 14:32:44 GMT</pubDate>
            <atom:updated>2018-11-14T14:32:44.616Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>1. Explain techniques used to avoid over-fitting in machine learning.</strong></p><p>Below are the techniques to prevent over-fitting problems:</p><p>- <strong>Use validation data set: </strong>Using the validation data set for evaluating the performance of classifier and determining the stop point where the performance for the validation data set starts to get worse, we may avoid over-fitting problem.</p><p>- <strong>Train with more data</strong>: As the training with more data, it tends to improve the error rate and their generalizability. However, too long training steps incur over fitting, and thereby we need to stop somewhere during training.</p><p>- <strong>Cross-validation: </strong>By using cross-validation, it allows us to tune hyper-parameter with only original training set. Therefore, we can keep the testing data set as a truly unseen data set for selecting the final model.</p><p>2. <strong>What bad effects are concerned for the imbalanced training data set in classification problems? How can we handle the imbalance problem in classification?</strong></p><p>- The bad effects concerned for imbalanced training set in classification problems is the following:</p><p>o The decision boundary established by model tends to be biased towards the majority class of training set.</p><p>o Therefore, the accuracy is meaningless.</p><p>- List below are some techniques for handling the imbalance problem in classification</p><p>o <strong>SMOTE technique:</strong> SMOTE technique creates synthetic samples (it is artificially created) from the minor class to make the data in balance. The algorithm selects two or more similar instances (using distance measure) and take the difference between them and multiply the difference with a random number between 0 and 1 to generate the new point (synthetic data).</p><p>o <strong>Random Under-Sampling: </strong>To balance the training data set, this technique removes the majority class of training data set.</p><p>o <strong>Random Over-Sampling: </strong>To balance the training data set, this technique adds more the minority class of training data set.</p><p>o <strong>Collect more data </strong>to balance the training set to fulfill the minority class data set</p><p>3. <strong>Explain why the cross-entropy function is useful as an error function in terms of probabilistic view.</strong></p><p>The usefulness of cross-entropy function is the range of output value from 0 to 1. Also, if the target output gets close to model output for all training inputs, then cross-entropy will tend toward zero — it implies that the neuron gets better at computing the target output for all training inputs. In other word, by using cross-entropy, the output vector represents a probability distribution, and cross-entropy indicates the distance between what the neural network believes this distribution should be.</p><p>4. <strong>What are the objectives of SVM learning? Express them in an optimization problem</strong></p><p>The main objectives of SVM in an optimization problem is that there are many possible separating hyper-planes for a given set of labeled training data, and therefore we have a choose an optimized one. In other word, in SVM, Optimized problem answers the question how to find the separating hyper-plane that not only separates the training data but also generalizes to new inferred data as well.</p><p>5. <strong>What are the assumptions made in Naïve Bayes classifiers?</strong></p><p>The assumption made in Naïve Bayes classifiers is that the features are independent of each other given the class.</p><p>6. Explain the characteristics of deep learning compared to the conventional neural networks.</p><p>· <strong>Conventional neural networks</strong></p><ul><li>Consists of 3 layers (input, hidden layer, and output)</li><li>Each node in a layer is interconnected to other nodes, and it has weight assigned to each edge</li><li>The input layer in network uses for receiving the data (probably in different format such as text, csv file, image, etc.)</li><li>The hidden layer is where all the calculation, computation and feature extraction take place.</li><li>The output layer is responsible for generating the desired output.</li></ul><p>· <strong>Deep learning</strong></p><ul><li>Consists of multiple hidden layers (input, N hidden layer, and output)</li><li>The hidden layer uses to transforms the input data into more abstract representation, and each hidden layer may take a unique role. For instance, hidden layer #1 uses to detect the edge of image, hidden layer #2 is used for detecting the general form that specified image and so on).</li><li>The series of layers, between input and output, identify the effective features from raw data during learning and therefore, it achieves better performance.</li><li>Carry out both feature extraction and learning simultaneously in a network.</li></ul><p>7. Explain why the dropout can be helpful in handling over-fitting for neural network</p><p>Dropout technique works by randomly selecting and removing neurons in a neural network while in training phase. By doing so, the random removal of neurons is ignored their coming and outgoing connections with the other nodes (including the same layer) while training the network and therefore, it reduces the likelihood of the over-fitting for neural network.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=edcc753dd9af" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The knowledge gained from experiences and books]]></title>
            <link>https://medium.com/@horndaneth/the-knowledge-gained-from-experiences-and-books-ccc8e170f309?source=rss-1dd1391a900d------2</link>
            <guid isPermaLink="false">https://medium.com/p/ccc8e170f309</guid>
            <dc:creator><![CDATA[Horn Daneth]]></dc:creator>
            <pubDate>Mon, 08 Oct 2018 14:06:11 GMT</pubDate>
            <atom:updated>2019-04-11T04:53:07.703Z</atom:updated>
            <content:encoded><![CDATA[<h3>Not everything contained in books</h3><p>As human-being, we gradually gain knowledge from day to day, but just some people have not realized that they had been changing. Two well-known bases, from which generally we acquire the knowledge, are books and experiences. As some people perspectives, both books and experiences are complementary needs to be a better quality of life. However, some people give different perspectives on the benefits or drawbacks between the books and experiences for some reasons.</p><p>One major difference between books and experiences is the amount of information. The experiences tend to give us more roughly information than the book since the book may probably not contain everything. Through experiences, the people deliberate on the real situations to which the specified problem. As a result, they are likely to get thorough information. Nonetheless, books may provide different aspects of problems. For instance, most of the books may cite contents from other books as their references, and thereby books should be consisting of the different aspects of the problems. Therefore, widely and thoroughly knowledge is an important combination for everyone.</p><p>Another difference between books and experiences is structural information. Due to the author preparation, the books may consist of structural knowledge. Whereas, the experiences are unstructured knowledge. Basically, the books always started from an introduction to a conclusion section. And also, before publishing any books, the authors usually check the coherence of contents. Furthermore, a book literally serves as some specified purposes from the authors. Being a well-written book, the authors should be focused on a specified problems domain to provide comprehensive and implicit information in the explicitly ordering to readers because the unordered issues may lead readers to have vague imaging in a book. In contrast, the experiences provide unordered knowledge which happened to the situation that people met. As a result, people like to read books because of their structural and comprehensive knowledge.</p><p>The differences between books and experiences are not only the amount and the structure of information but also the time to be gained knowledge. Because experiences are something about ourselves, we may not have any signals to what the problem is, and therefore it requires some time to extract the implicit knowledge from what we encountered. On the other hand, the books may guide us about scenarios of the issues and the way to address the problems. Such that we promptly gain the knowledge from the book. Also, the experiences are tough activities to turn into the knowledge as many people could probably not observe that they are gaining the knowledge from it. Lastly, Experiences are costly actions to gain in something new.</p><p>In sum up, books and experiences are intrinsic activities in our daily basis. Despite their advantages, it is still remaining some challenges for us. However, we should take some advantages from the books, and we combine with our experiences for a better life and career.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ccc8e170f309" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>