A survey of Automated Data Augmentation: Learning to Compose, Mix and Generate

PhD Qualifying Examination


Title: "A survey of Automated Data Augmentation: Learning to Compose, Mix 
and Generate"

by

Mr. Tsz Him CHEUNG


Abstract:

Data Augmentation is an effective way to improve the generalization 
capability of modern deep learning models. However, the underlying 
augmentation methods mostly rely on handcrafted operations, like 
horizontal flipping and randomized cropping for image data. Moreover, an 
augmentation policy useful to one dataset may not transfer well to other 
datasets. Therefore, Automated Data Augmentation (AutoDA) methods are 
proposed to automate the process of searching for optimal augmentation 
policies. In this survey, we will explain the motivation and challenges of 
AutoDA, review the recent developments of AutoDA methods, analyze their 
effectiveness and efficiency, demonstrate the application of AutoDA on 
various machine learning tasks and data modalities as well as provide 
future insights to enrich and extend the existing approaches.


Date:			Monday, 19 July 2021

Time:                  	3:00pm - 5:00pm

Zoom meeting:
https://hkust.zoom.us/j/95877658018?pwd=aWlpeHI1UHhQMmNmVVBXTEtocW1wUT09

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


**** ALL are Welcome ****