Explore adversarial methods for text in this comprehensive lecture from CMU's Advanced NLP course. Delve into generative adversarial networks, examining their applications in natural language processing. Learn about the strategic placement of adversaries in features versus outputs, and understand the challenges of implementing GANs with discrete outputs. Investigate techniques for handling adversaries on discrete inputs, gaining insights into cutting-edge NLP research. Cover topics such as distribution matching, discriminator training, stabilization tricks, and unsupervised style transfer. Enhance your understanding of advanced NLP concepts through this in-depth exploration of adversarial techniques in text processing.
Overview
Syllabus
Introduction
Adversarial Methods
generative adversarial networks
No generative models
Basic Paradigm
Training Method
Distribution Matching
Pseudocode
Why
Problems
Applications
Learning Methods
Discriminators
Comparing two outputs
Training a discriminator
Stabilization tricks
Discriminator over results
Adversarial Feature Learning
MultiTask Learning
Professor Forced
Unsupervised Style Transfer
Unsupervised Alignment
Taught by
Graham Neubig
Related Courses
-
CMU Advanced NLP 2021 - Adversarial Learning
-
Neural Nets for NLP 2021 - Adversarial Methods
-
Neural Nets for NLP 2017 - Adversarial Learning
-
Deep Learning for Text with PyTorch
-
Neural Nets for NLP 2017 - Multilingual and Multitask Learning
-
CMU Neural Nets for NLP 2020 - Multitask and Multilingual Learning
