A Literature Survey on Active learning: Active Learning with Complex Structures

PhD Qualifying Examination


Title: "A Literature Survey on Active learning: Active Learning with
Complex Structures"

Mr. Yi ZHEN


Abstract:

Many supervised learning tasks need large amounts of labeled data.
However, labeling data is always difficult, expensive and
time-consuming, especially when domain knowledge is hard to obtain.
Active learning, which reduces the labeling cost without loss of model
quality, has recently become a hot research topic in the area of
machine learning, information retrieval, data mining, etc. As the most
interesting and realistic scenario of active learning, pool-based
active learning has been widely studied in lots of applications. This
paper presents a survey on existing algorithms for pool-based active
learning. Firstly, we introduce some backgrounds of active learning
and divide existing active learning algorithms into two categories,
namely active learning with simple structures (ALSS) and active
learning with complex structures (ALCS), according to their
assumptions and settings. Then, algorithms of ALSS and ALCS are
reviewed in two consecutive sections respectively. After that, some
works in related areas will be introduced briefly. Finally, we will
discuss some possible topics for future research.


Date:     		Thursday, 8 October 2009

Time:                   10:00am - 12:00noon

Venue:                  Room 2404
 			lifts 17/18

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


**** ALL are Welcome ****