Explore variational autoencoders in this 43-minute lecture from Northeastern University's CS 7150 Deep Learning course. Delve into generative models, plain autoencoders, variational and evidence lower bounds, variational autoencoder architecture, and stochastic optimization techniques. Access comprehensive notes and references, including works by Kingma and Welling, to deepen your understanding of this advanced machine learning topic. Gain insights into latent codes, likelihood optimization, noisy observation models, KL divergence, regularization, and multistep optimization processes in the context of variational autoencoders.
Overview
Syllabus
Introduction
What is agenerative model
Latent code
Autoencoders
likelihood optimization
generative model
nosy observation model
setup
lower bound
KL divergence
Regularization
Maximizing the Lower Bound
Multistep Optimization
Variational Autoencoders
Stochastic Gradient Optimization
Key Points
Taught by
Paul Hand
