Jekyll2021-12-24T11:51:30-06:00https://theguywithblacktie.github.io/kernel/feed.xmlNLP in day, CV by night.\ Documenting my learnings of Machine Learning.\ Motto: Be Limitless 2021-07-20T00:00:00-05:002021-07-20T00:00:00-05:00https://theguywithblacktie.github.io/kernel/deep%20learning/2021/07/20/BLEU

BLEU, or Bilingual Evaluation Understudy, is a performance metric score for comparing the machine translations and human created reference translations for the same source sentence. This metric was introduced in Kishore Papineni, et al. in their 2002 paper “BLEU: a Method for Automatic Evaluation of Machine Translation”

2021-06-30T00:00:00-05:002021-06-30T00:00:00-05:00https://theguywithblacktie.github.io/kernel/deep%20learning/2021/06/30/GAN

We have come across various articles and posts about AI being capable of producing human like speech or generating images of non-existing people that are difficult to distinguish from real-life existence. These AI systems are built upon generative adversarial networks (GANs) - which Facebook AI research director Yann LeCun called “most interesting idea in last 10 years in ML” GANs were introduced in a paper by Ian Goodfellow and other researchers at the University of Montreal, including Yoshua Bengio, in 2014.

2021-06-06T00:00:00-05:002021-06-06T00:00:00-05:00https://theguywithblacktie.github.io/kernel/machine%20learning/2021/06/06/SVM

In Machine Learning, out of the many available supervised classification algorithms, Support Vector Machine (SVM) is one of the easiest algorithm and yet sometimes it becomes difficult to grasp due to its little nuances. In this blog, I try to lay down my notes on unboxing the SVM blackbox and make it easy to understand in detail. This blog focusses on entirety of SVM, from its introduction to classification and kernel tricks.

2021-05-20T00:00:00-05:002021-05-20T00:00:00-05:00https://theguywithblacktie.github.io/kernel/machine%20learning/pytorch/2021/05/20/cross-entropy-loss

[Updated on 03-12-2021: Fixed the Focal Loss Code]

2021-05-14T00:00:00-05:002021-05-14T00:00:00-05:00https://theguywithblacktie.github.io/kernel/pytorch/2021/05/14/vary-layers-pytorch

Consider following code block that defines a fixed 2-layer neural network. Imagine a scenario, where the network has huge number of layers, and typing out each layer manually is just not feasible. An even more notable scenario is when the number of layers of network are not fixed and it depends on some other conigurations. This article deals with these scenarios and lays out solution.