Neural Nets for NLP 2021 - Multilingual Learning

Explore multilingual learning in neural networks for NLP through this lecture from CMU's CS 11-747 course. Delve into topics such as generative vs. discriminative models, deterministic vs. random variables, variational autoencoders, and handling discrete latent variables. Examine practical examples of variational autoencoders in NLP applications. Gain insights into multilingual models, heuristic sampling, parameter sharing, and adapters. Investigate crosslingual word recovery, benchmarks, and transfer learning techniques. Learn about annotation projection, language overlap, syntax considerations, and active learning approaches for multilingual NLP tasks.

Intro
Multilingual Learning
Flowchart
Multilingual Models
Heuristic Sampling
Balance Data Directly
Share Parameters
Adapters
Multilingual Pretraining
Crosslingual Word Recovery
Benchmarks
Crosslingual Transfer Learning
Pretrained and Finetuned Paradigm
Annotation Projection
Transfer
Language Overlap
Language Syntax
Active Learning
Uncertainty
Sampling Criteria