Abstract: Manipulating videos and photographs to edit artifacts has been in practice for quite a long time. If you have seen the movie Fast and Furious 7, chances are you did not even notice how seamlessly the scenes Paul Walker were added to the movie. Along the same lines is a Buzzfeed video of former US president Barack Obama, where he says "Killmonger was right". The former is the result of painstaking manual work done using complex visual effects/CGI. The latter, on the other hand, is the result of a technology called deepfakes. A portmanteau of the words deep learning and fake, deepfake is a broad term used to describe AI-enabled technology that is used to generate the examples we discussed. By completing this workshop, you will develop an understanding of the deepfakes landscape and deepfakes workflow along with hands-on guide to train a very basic deepfake setup of your own
Part I: Deepfakes Overview
Familiarize yourself with the overall deepfakes landscape by learning about different use-cases (both productive and malicious ones). Understand different modes of operation along with some key feature sets that are leveraged by different methods
Part II: DeepFakes Workflow
Deepfakes is evolving like all other sub-fields in the AI universe. In this session we will focus on understanding a general workflow and some common architectures which are leveraged widely to train deepfakes models
Part III: Train your own DeepFakes
Let’s leverage our understanding to develop a very basic deep-fakes setup. We will make use of Jupyter notebooks and publicly available datasets for this step and understand how effectively we can leverage simple methods/architectures to great impact.
Python, TensorFlow2, Basics of Machine Learning and Deep Learning
Bio: Raghav Bali is a Senior Data Scientist at Optum(United HealthGroup), one of the world’s largest health care organizations. With about a decade’s experience working across Fortune 500 organizations such as Intel and American Express, his work involves research & development of enterprise-level solutions based on Machine Learning, Deep Learning, and Natural Language Processing for real-world use-cases. Raghav has published multiple peer-reviewed papers, has authored over 7 books, and is a co-inventor of multiple patents in the areas of machine learning, deep learning, and natural language processing.