Walkthrough of Deep Learning Using Various Open Source Frameworks
This session covers various dimensions of deep learning through variation of examples with the help of diverse technologies. Talk-1: The first session introduces deep learning and talks about Recurrent Neural Networks (RNN), LSTM. The session walks through the deep learning framework Theano. Then a natural language processing problem is solved using the Theano framework. Talk-2: The second session briefs about Convolutional Neural Networks (CNNs) which help in solving image recognition problem. The session walks through the deep learning framework Keras using CNTK as backend. An image recognition problem is then solved and the intermediate visualizations are demonstrated using Jupyter Notebook. Talk-3: This session will familiarize audience with Deep Reinforcement Learning concepts. The tools used will be Keras with Tensorflow and Open AI Gym, use of which will be demonstrated through code walkthrough of a simple game. Resources for self learning will be provided.