Machine Learning and
Content-Based Information Retrieval and Multimedia Database
Dr. John Gan
Multimedia information are now widely available in various areas.
It becomes cheaper and cheaper to acquire and store multimedia
information. However, multimedia retrieval is still a bottleneck.
Almost all the multimedia information available from the Internet is
still retrieved by matching manually generated annotations. The key
issues in this topic include how to automatically extract effective
content indexes (feature vectors) from multimedia information and
how to match a content query with a huge amount of stored content
indexes. We are investigating efficient multimedia indexing
structures and effective criteria for content-based similarity
search. We are also interested in multimedia database systems.