I like to read and talk about Math, Machine Learning and Artifical Intelligence. This blog is part of my effort to keep learning more and evolving as I move forward.

Posts

  • Loss Functions Part 2

    In this part of the multi-part series on the loss functions we'll be taking a look at MSE, MAE, Huber Loss, Hinge Loss, and Triplet Loss. We'll also look at the code for these Loss functions in PyTorch and some examples of how to use them

    Aug 10, 2021

  • Loss Functions Part 1

    In this part of the multi-part series on Loss Functions we'll look at the concepts like Information Entropy, Cross Entropy, KL-Divergence, how they are linked to each other and how to code them in Python

    Aug 4, 2021

  • Optimization Notes

    In this post we'll go over Gradient based optimization. Understanding gradient based optimization methiods is very important if someone needs to become an expert in Deep Learning. I'll start by a quick refresher on univariate and multivariate optimization followed by a brief overview of some of the Gradient based optimization methods.

    Jul 27, 2021

  • A comparison of Pets classifier using vanilla PyTorch and Fast.ai

    Comparing fine tuning of a RestNet34 based Pets classifier using vanilla PyTorch code with the one written using Fast.ai. The purpose of this blog is to demonstrate how `fastai` can really help one get started with `deep learning` and also provide right abstractions and encapsulation that helps one focus more on the research and modeling than to spend our time in boiler plate code. `fastai` is amazing at abstracting away a lot of inner details that we should focus on but not when we're starting out or when we're prototyping our research. The post below shows that what usually requires writing custom classes for creating `data-splitters`, `datasets`, `trainers` without using a library like `fast.ai` can be accomplished in very few lines of code with `fast.ai`. I start out by writing that code assuming we don't have access to `fast.ai` and then I go and surprise myself by showing how easy it is to perform the same task using `fast.ai`.

    Jul 20, 2021