mindis/cs236_dgm

🦍 Stanford CS236 : Deep Generative Models

CS236 : Deep Generative Models

CS236 Fall 2018, The "IAN" class of Stanford. Generative Models or "GANS" in the spotlight, here I begin my CS236 journey. Though I didn't enroll in the class, I used my stanford email to set up my lab (Google cloud coupons). The course is new, "first taught" this quarter, lets keep learning.

Course

Generative models are widely used in many subfields of AI and Machine Learning. Recent advances in parameterizing these models using deep neural networks, combined with progress in stochastic optimization methods, have enabled scalable modeling of complex, high-dimensional data including images, text, and speech. In this course, we will study the probabilistic foundations and learning algorithms for deep generative models, including Variational Autoencoders (VAE), Generative Adversarial Networks (GAN), autoregressive models, and normalizing flow models. The course will also discuss application areas that have benefitted from deep generative models, including computer vision, speech and natural language processing, graph mining, and reinforcement learning.

• Grading : Homeworks (15% x 3 = 45%) + Midterm: 15% + Course Project 40%

Notes @github

⌘ Book - Deep Learning by Ian Goodfellow, Yoshua Bengio, Aaron Courville. | [pdf]

🎄 Homeworks

☃ Homework 1 : Starter Zip : Solution
☃ Homework 2 : Starter Zip : Solution
☃ Homework 3 : Starter Zip : Solution

Course :

♞ INTRODUCTION

𓁅 Introduction and Background (slides 1, slides 2)

♞ AUTOREGRESSIVE MODELS

𓁅 Autoregressive Models (slides 3, slides 4)

♞ VARIATIONAL AUTOENCODERS

𓁅 Variational Autoencoders (slides 5, slides 6)

♞ NORMALIZING FLOW MODELS

𓁅 Normalizing Flow Models (slides 7, slides 8)

♞ GENERATIVE ADVERSARIAL NETWORKS

𓁅 Generative Adversarial Networks (slides 9, slides 10)

• Guest lecture 1: Tengyu Ma, Evaluation of Generative Models on Wednesday (slides 11)
  
• Combining generative model variants (slides 12), Energy-based models (slides 13)
  
• Guest lecture 2: Diederik P. Kingma, Discreteness in Latent Variable Modeling on Wednesday (slides 14)
  
• Applications: Vision, Speech, Language, Graphs, Reinforcement learning
  
• Generative Adversarial Imitation Learning (GAIL)
  
• Introduction to PyTorch
  
• Graphite : Iterative Generative Modelling of Graphs
  

Additional Reading: Surveys and tutorials

• Papers by Guest Lecturer -Diederik- Auto-Encoding Variational Bayes, Improving Variational Inference with Inverse Autoregressive Flow and Glow- Generative Flow with Invertible 1x1 Convolutions
• Tutorial on Generative Adversarial Networks. Computer Vision and Pattern Recognition, June 2018.
• Tutorial on Deep Generative Models Shakir Mohamed and Danilo Rezende. Uncertainty in Artificial Intelligence, July 2017.
• Tutorial on Generative Adversarial Networks Ian Goodfellow. Neural Information Processing Systems, December 2016
• Learning deep generative models. Ruslan Salakhutdinov. Annual Review of Statistics and Its Application, April 2015.
• Generative Models from OpenAI
• Collection of generative models, e.g. GAN, VAE in Pytorch and Tensorflow. : wiseodd.github.io
• Deep.Generative.Models.cs.nyu.edu
• Aweome GANs paper, Application of GANs, Generative Models
• GANs - The Story so Far

YouTube

• MIT 6.S191 (2018): Deep Generative Modeling
• Stanford CS231n - Generative Models
• Siraj Raval - Generative Models - The Math of Intelligence #8
• MIT 6.S191 (2019): Deep Generative Modeling
• Generating Pokemon with a Generative Adversarial Network
• DeepMind - From Generative Models to Generative Agents - Koray Kavukcuoglu

✡ Exam : Fall@2018-Mid | Collected from public resources

FINAL PROJECT

The Final Project is important and here are the resources - Project Guidelines, Project Proposal Guidelines, Final Report Guidelines, Project Examples and the Nips format to write the Final Project Paper in LaTeX. I ended up doing " ".