<?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 Lahiri Mahasaya Bestha on Medium]]></title>
        <description><![CDATA[Stories by Lahiri Mahasaya Bestha on Medium]]></description>
        <link>https://medium.com/@lahirimahasaya628?source=rss-485fa5ccf8a0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*rDN-n-t_27n4yYNl905VgA.jpeg</url>
            <title>Stories by Lahiri Mahasaya Bestha on Medium</title>
            <link>https://medium.com/@lahirimahasaya628?source=rss-485fa5ccf8a0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 15:18:16 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@lahirimahasaya628/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[Synthetic Data Generation]]></title>
            <link>https://medium.com/@lahirimahasaya628/synthetic-data-generation-c3e09dc097?source=rss-485fa5ccf8a0------2</link>
            <guid isPermaLink="false">https://medium.com/p/c3e09dc097</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[gans]]></category>
            <category><![CDATA[python-programming]]></category>
            <dc:creator><![CDATA[Lahiri Mahasaya Bestha]]></dc:creator>
            <pubDate>Sun, 09 Jan 2022 17:25:08 GMT</pubDate>
            <atom:updated>2022-01-10T07:55:08.013Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Introduction</strong></h3><p>As we know data is key for everything, But finding data is a strenuous task 🤔. So rather than finding data why can’t we create our own. 😉</p><p><strong><em>Now let me give an example, In India normally such as Pan Card, Aadhar Card data is not readily available due to security reasons, So in this scenario, we need to generate this kind of data synthetically.</em></strong> 😅</p><p>So in this blog, I will explain “ How at SandLogic data is created by using Generative adversarial networks”. I will be using an example of a synthetically generated Pan Card.</p><p><strong><em>What are Generative Adversarial Networks?</em></strong></p><blockquote>The generator takes in random numbers and returns an image.</blockquote><blockquote>This generated image is fed into the discriminator alongside a stream of images taken from the actual, ground-truth dataset.</blockquote><blockquote>The discriminator takes in both real and fake images and returns probabilities, a number between 0 and 1, with 1 representing a prediction of real and 0 representing fake.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*jB5LSEh9I_8FrSin.png" /><figcaption>Generative Adversarial Networks</figcaption></figure><h3><strong>Steps</strong></h3><p>This is the pipeline used for the generation of the Synthetic data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qAGHNrOymzTVjnBV_ewG1w.jpeg" /><figcaption><strong>Pipeline</strong></figcaption></figure><p>The First step is mainly to generate the Faces and As we know that we cannot randomly use others images so we use the GAN’s to generate. We have a custom in-house face generation GAN model where we can select a variety of faces that can be generated out of random noise.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/128/1*cCE8AFKlqaoDAwN64pC3xg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/128/1*0REjfvH52u9K6WT8P8XtNw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/128/1*9hQ6XwIPLGBIz8_n3LwB3g.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/128/1*jp_TgKfU9HVlwjj4o9Eeow.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/128/1*fwv79JklbP8TduOK7iG_NA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/128/1*jmEevcwTapBzIwv8A448aw.png" /></figure><p><strong>These are some synthetically generated faces using Our model.</strong></p><p>Secondly, we use different templates such as PAN Card, Aadhar card etc.., We merge the faces Generated and the templates using image processing techniques.</p><p>Thirdly, The Pan Card will be having details such as the Name, Father Name, Date of Birth PAN Card number, Signature and QR Code which are automatically generated by our pipeline.</p><p>Before putting the random name we are checking the gender by using our custom Gender Recognition model and then we are adding the name based on its prediction. As we know that the father’s name should be always male so we are generating the male names in that field directly.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/250/1*pileGDHigOoriIXA7EpAag.png" /></figure><p>And in the PAN Card number, the 5th letter is the first name’s first letter we are also handling that.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/284/1*OfUQVeKQQB7tf8osZNbLRg.png" /></figure><p><em>In the above example, the 5th letter is G which is from Grover.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/234/1*b3Ug3roJlyzLEUx8UxBDdQ.png" /></figure><p><em>And lastly, we are generating a Signature as well as the QR code.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/203/1*4sTMLZzTHQLm9BPJhNaNgw.png" /></figure><p><em>QR Code has the details like PAN Number and Name and Fathers Name and date of birth in it.</em></p><p>After the end-to-end pipeline, the final output is shown below</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/1*kkHLBJBhqK01fK1A3QtfXA.png" /><figcaption><strong>Final Output</strong></figcaption></figure><h3><strong>Conclusion</strong></h3><p>By this, we can conclude that rather than wasting time finding data we can create our own synthetic data.</p><blockquote><em>We at SandLogic believe in building strong fundamentals, one among them is to use “Efficient Engineering” to solve the most complex problems — “Keep it simple”.</em></blockquote><p>Talk to us at <a href="https://www.linkedin.com/company/sandlogic">Linked In</a> if you have any queries or a business problem to discuss. We are always ready to help.</p><p>SandLogic Technologies Pvt. Ltd. © 2021. All rights reserved.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c3e09dc097" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>