Explore parsing with dynamic programs in this lecture from CMU's Neural Networks for NLP course. Delve into graph-based parsing, minimum spanning tree parsing, and structured training techniques. Learn about dynamic programming methods for phrase structure parsing and reranking approaches. Examine algorithms like Tulio Edmunds and IceNurse, and understand the transition from traditional to neural models. Investigate global probabilistic training, CKY and Viterbi algorithms, and Conditional Random Fields (CRFs) for parsing. Gain insights into neural CRFs, structured inference, and recursive neural networks for parsing tasks.
Overview
Syllabus
Introduction
Linguistic Structure
Dynamic Programming Based Models
Minimum Spanning Tree
Graph Based vs Transition Based
Tulio Edmunds Algorithm
IceNurse Algorithm
Quiz
Before Neural Nets
Higher Order Dependency Parsing
Neural Models
Motivation
Model
Example
Global probabilistic training
Code example
Algorithms
Phrase Structures
Parsing vs Tagging
Hyper Graph Edges
Scoring Edges
CKY Algorithm
Viterbi Algorithm
Over Graphs
CRF
CRF Example
CRF Over Trees
Neural CRF
Inference
Parsing
Structured Inference
Recursive Neural Networks
Rear Inking
Rear Inking Results
Next Time
Taught by
Graham Neubig
Related Courses
-
Neural Nets for NLP 2017 - Neural Semantic Parsing
-
Neural Nets for NLP - Transition-based Parsing
-
Neural Nets for NLP 2017 - Transition-Based Dependency Parsing
-
CMU Neural Nets for NLP 2017 - Advanced Search Algorithms
-
Neural Nets for NLP 2020 - Generating Trees Incrementally
-
Neural Nets for NLP 2018 - Neural Semantic Parsing
