A Survey of Neural Language Models

PhD Qualifying Examination


Title: "A Survey of Neural Language Models"

by

Miss Ziqian ZENG


Abstract:

A language model is a probability distribution over a word sequence. 
Language models are very useful in a broad range of applications, such as 
speech recognition, optical character recognition, machine translation, 
generation, and context sensitive spell correction. Neural language models 
(NLMs) are considered to show better performance than traditional language 
models. In this survey, we will introduce two kinds of neural language 
models: feed-forward neural language models and recurrent neural language 
models. We also introduce three optimization techniques to train a neural 
language model, namely, importance sampling, noise-contrastive estimation, 
and hierarchical Softmax. Finally, We show some possible research problems 
in neural language models, such as out-of- vocabulary, using feature 
context and NLM adaptation.


Date:			Wednesday, 13 June 2018

Time:                  	3:00pm - 5:00pm

Venue:                  Room 5560
                         Lifts 27/28

Committee Members:	Dr. Yangqiu Song (Supervisor)
 			Dr. Wei Wang (Chairperson)
 			Prof. James Kwok
 			Prof. Nevin Zhang


**** ALL are Welcome ****