Description:

Talk Abstract

This talk aims to introduce Facebook's Deep Learning library - PyTorch. PyTorch is increasing becoming popular due to it's powerful features and shallow learning curve, making it accessible and easy to use to a larger community. This talk discusses about the 'Autograd' package, which is central to all neural networks in PyTorch. It also entails details about the different features and functionalities of PyTorch as well as equip the audience on how to create simple and complex Neural Networks in PyTorch. PyTorch helps to create dynamic computation graphs that allow you to change how the network behaves on the fly unlike static computation graphs. It offers modularity which enhances the ability to debug or see within the network.

Outline of the Talk

The talk will be broadly divided into 3 broad parts.

Part 1 will be an Introduction to PyTorch. This part will focus on the use and need for PyTorch as a deep learning framework. This will be followed by instructions on how to setup PyTorch and a look at the basic building blocks behind the framework.

Part 2 will dive more into the features of PyTorch, mainly it's AutoGrad package which lies at the heart of all Neural Networks created in PyTorch and PyTorch's ability to create dynamic computational graphs as opposed to the static computational graphs offered by some of it's counterparts (such as TensorFlow and Caffe).

Part 3 will be a more 'hands on' part where the talk will focus on how to create and build simple as well as complex neural networks (such as Convolutional Neural Networks) with the framework.

Prerequisites:

1. A basic understanding of how Neural Networks work would be beneficial.
2. Some knowledge about Numpy.

Content URLs:

1. https://pytorch.org/docs/stable/index.html
2. Slides (https://slides.com/rahulbaboota/deck)

Speaker Info:

I am Rahul Baboota, a 3rd Year Undergraduate in India studying Computer Science and Engineering. I have an avid interest in the domain of Data Science, Machine Learning and Deep Learning. I have worked at various Data Science and Machine Learning based startups and labs. In my freshmen year, I worked at a data journalism startup to create and analyze smart data stories. I was also a part of a project funded by the Government of India for the development of a social media based analytics tool for the analysis of healthcare and nutrition in India. I am currently working at the Center for Artificial Intelligence at IIITD in the Autonomous Vehicle Lab 'Swarath'.

Speaker Links:

1. https://www.linkedin.com/in/rahulbaboota/
2. https://github.com/RahulBaboota