A Survey on Neural Network-Based Word Embedding

PhD Qualifying Examination


Title: "A Survey on Neural Network-Based Word Embedding"

by

Miss Lanqing XUE


Abstract:

The objective of word embedding is to associate each word with a dense, 
low-dimensional, and real-valued vector. Word embedding is one of the most 
successful applications of unsupervised learning method. It has been applied to 
multiple tasks in Natural Language Processing (NLP), such as part-of-speech 
tagging (POS), named entity recognition (NER), semantic-role labeling (SRL), 
chunking and parsing in dialogue systems, question and answering (QA), machine 
translation and sentiment analysis. It is also commonly used in Computer Vision 
(CV) for tasks such as image captioning. With the advances of neural network 
algorithms in recent decades, neural network-based word embeddings are becoming 
more syntactically and semantically meaningful than traditional word 
representations, and these good embeddings improve model performances over all 
tasks. Thus, word embedding plays an increasingly significant role both in NLP 
and CV. This survey gives an overview of the developments of neural 
network-based word embedding, proposes a categorization of all major algotithms 
based on the types of word co-occurrences they use, highlights challenges in 
current researches and points out possible future directions.


Date:			Thursday, 5 July 2018

Time:                  	10:00am - 12:00noon

Venue:                  Room 5560
                         Lifts 27/28

Committee Members:	Prof. Nevin Zhang (Supervisor)
 			Dr. Brian Mak (Chairperson)
 			Dr. Yangqiu Song
 			Dr. Raymond Wong


**** ALL are Welcome ****