Build a Deep Convolutional Generative Adversarial Networks using Pytorch

Akash M (~akash94)


Description:

Deep learning has been proven in recent years to be an extremely useful tool for discriminative tasks. Through layers of linear transforms combined with nonlinearities, these systems learn to transform their input into an ideal representation across which we can draw clear decision boundaries. Systems with discriminative deep-learning models at their core now yield state-of-the-art results in image classification, speech recognition, and many other popular tasks of modern artificial intelligence. This talk aims to discuss the basic concepts and working of Generative Adversarial Networks and show a live example of how to build a Deep Convolutional Generative Adversarial Networks.

Outline -

  • Self Introduction - I will be giving welcome notes and introduce myself to everyone.
  • Introduction to and working of GAN - A brief introduction about GAN and visual presentation of GAN will be shown.
  • What is DCGAN - A brief introduction about DCGAN will be given.
  • Walk through to the Face Generation project - A walk through to the code and explaining about generators and discriminators and their concepts and get the results
  • Conclusion : Where to go next - Will be telling about how to improve the code to get greater results.
  • Sharing resources.

Prerequisites:

Preferred to have -

      1. Basic knowledge of Linear Algebra
      2. Basic background in Python Programming
      3. Basic knowledge of Neural Network and Convolution Neural Network

Content URLs:

The Slide is available here

Speaker Info:

Akash is a Computer Science major at Sathyabama Institute of Science and Technology. His major interest is Artificial intelligence and Robotics and he is also a cyber security enthusiast. He is a technophile love to learn and explore new technologies. He have worked on various Deep learning, Web and Android projects and have participated in some CTF competitions and hackathons.

Speaker Links:

Id: 1127
Section: Data Science, Machine Learning and AI
Type: Talks
Target Audience: Intermediate
Last Updated: