IIT Madras · Computer Science
Deep Learning - Part 2
Prof. Mitesh M. Khapra · 37 free video lectures
37 video lectures. Topics: Recap of Probability Theory; Why are we interested in Joint Distributions; How do we represent a joint distribution; Can we represent the joint distribution more compactly; Can we use a graph to represent a joint distribution; Different types of reasoning encoded in a…
▶ Play on Open Lecture HallCourse page ↗
Lectures
- Recap of Probability Theory
- Why are we interested in Joint Distributions
- How do we represent a joint distribution
- Can we represent the joint distribution more compactly
- Can we use a graph to represent a joint distribution
- Different types of reasoning encoded in a Bayesian Network
- Independencies encoded by a Bayesian Network(Case 1: Node and it's parents)
- Independencies encoded by a Bayesian Network(Case 2: Node and it's non-parents)
- Independencies encoded by a Bayesian Network(Case 3: Node and it's descendants)
- Bayesian Networks : Formal Semantics
- I-Maps
- Markov Networks: Motivation
- Factors in Markov Network
- Local Independencies in a Markov Network
- Joint Distributions
- The concept of a latent variable
- Restricted Boltzmann Machines
- RBMs as Stochastic Neural Networks
- Unsupervised Learning with RBMs
- Computing the gradient of the log likelihood
- Motivation for Sampling
- Motivation for Sampling - Part - 02
- Markov Chains
- Why de we care about Markov Chains ?
- Setting up a Markov Chain for RBMs
- Training RBMs Using Gibbs Sampling
- Training RBMS Using Contrastive Divergence
- Revisiting Autoencoders
- Variational Autoencoders: The Neural Network Perspective
- Variational Autoencoders: The Graphical model perspective
- Neural Autoregressive Density Estimator
- Masked Autoencoder Density Estimator (MADE)
- Generative Adversarial Networks - The Intuition
- Generative Adversarial Networks - Architecture
- Generative Adversarial Networks - The Math Behind it
- Generative Adversarial Networks - Some Cool Stuff and Applications
- Bringing it all together (the deep generative summary)
From NPTEL (IITs and IISc) · CC BY-NC-SA. More from IIT Madras.
Open Lecture Hall is free, with no ads and no sign-up. It is not run by or connected to any university; the lectures are the universities' own free recordings, played from where they host them. About & rights