Explore adversarial learning in natural language processing through this advanced lecture from CMU's CS 11-711 course. Delve into generative adversarial networks, examining their applications in both feature and output spaces. Investigate the challenges of applying GANs to discrete outputs and learn about adversarial techniques for discrete inputs. Gain insights into distribution matching, image generation, and unsupervised style transfer in language processing. Enhance your understanding of advanced NLP concepts and their practical implementations in this comprehensive 81-minute session led by Graham Neubig.
Overview
Syllabus
Intro
Adversarial Methods
generative adversarial networks
nonlatent models
ML vs GAN
Basic Paradigm
Loss Function
Distribution Matching
Distribution Matching Pseudocode
Why are Gans good
Image Generation
Problems
Classes
Discriminators
Questions
Discrete choices
Domain and variant representations
Language variant representations
Unsupervised style transfer
Taught by
Graham Neubig
