A Survey on Testing of Deep Learning System

PhD Qualifying Examination


Title: "A Survey on Testing of Deep Learning System"

by

Mr. Yongqiang TIAN


Abstract:

Nowadays, it has been well realized that the failures in deep learning 
systems could bring serious accidents to human’s life. Before deployment, 
comprehensive testing of deep learning systems is necessary to validate 
the correctness and harmlessness of its functionality. However, existing 
techniques in software testing may not be easily applied to deep learning 
systems, since the deep learning systems are different from conventional 
software in many aspects. Till now, a lot of researchers have devoted 
their efforts to enhance the testing of deep learning models. In this 
survey, we reviewed a collection of representative works in the testing of 
deep learning systems. We classified these studies into two categories, 
i.e., code-level testing and model-level testing, according to the 
components of deep learning systems they targeted on. For each category, 
we introduced the representative works, summarized their contribution, and 
pointed out their limitations. We also presented some research 
opportunities that are worthy of exploration in the future.


Date:			Wednesday, 26 June 2019

Time:                  	2:00pm - 4:00pm

Venue:                  Room 3494
                         Lifts 25/26

Committee Members:	Prof. Shing-Chi Cheung (Supervisor)
 			Dr. Yangqiu Song (Chairperson)
 			Prof. Fangzhen Lin
 			Dr. Raymond Wong


**** ALL are Welcome ****