Multi-task Learning and Its Applications in Automatic Speech Recognition: A Survey

PhD Qualifying Examination


Title: "Multi-task Learning and Its Applications in Automatic Speech 
Recognition: A Survey"

by

Mr. Yingke ZHU


Abstract:

Multi-task learning is a learning mechanism that aims at improving 
generalization performance by using the domain knowledge contained in related 
tasks. It achieves this by learning tasks in parallel while using a shared 
representation. A common set-up is that there are multiple related tasks for 
which we want to get better performance by learning them simultaneously. 
However, simply assuming relatedness among tasks and learning them together can 
be detrimental. It is therefore important to capture relationships between 
tasks. This survey reviews prior works on multi-task learning and relationship 
modeling. In the last part, applications and potential research problems of 
multi-task learning in speech recognition are discussed.


Date:			Tuesday, 26 January 2016

Time:                  	10:00am - 12:00noon

Venue:                  Room 3584
                         Lifts 27/28

Committee Members:	Dr. Brian Mak (Supervisor)
 			Dr. Raymond Wong (Chairperson)
 			Prof. Dit-Yan Yeung
 			Prof. Nevin Zhang


**** ALL are Welcome ****