	News & Achievements

	Events

	Video

	Highlights

	Contact & Direction

AREAS

Computational Intelligence Networking and Systems Pattern Recognition and Machine Learning Databases and Information Management

Computational Intelligence

We focus on multi-agent autonomy-oriented computing, web intelligence and intelligent user interfaces. In particular, our research aims to develop intelligent algorithms and systems which can analyze, explain and self-organize to tackle real-world complex problems caused by the current ubiquity of diverse types of explicitly/implicitly linked digital contents, and the complexity of the underlying behaviors for systems and environments found in domains including education, public health and sustainability studies.

Faculty Involved:

Funded Research and Consultancy Projects in the Past Few Years:

A Critiquing System for Enhancing Writing Competency Using Automatic Text Analysis
	Staff	Dr. CHEUNG, William Kwok Wai (Project Investigator)
	Abstract	To build a web supported learning system with an automatic text analysis module in it, so that (1) teachers can monitor and manage individual student’s writing processes, and (2) students can submit drafts during their writing processes and receive just-in-time subtheme suggestions (critiques) from the system.

A Model-based Approach to Distributed Data Mining with Privacy Preservation
	Staff	Dr. CHEUNG, William Kwok Wai (Supervisor of PhD Student Mr. ZHANG, Xiaofeng)
	Abstract	To investigate a principled way of learning global statistical models (for clustering and manifold discovery) from physically distributed local data subsets. Research issues include deriving a proper local data abstractions, learning global generative models from the local abstractions, and enabling negotiation among the local data sources for optimizing the global model accuracy with a minimum cost.

Non-negative Matrix Factorization Based Belief Compression and Clustering for Solving POMDPs
	Staff	Dr. CHEUNG, William Kwok Wai (Supervisor of PhD Student Ms. LI, Xin)
	Abstract	The main objective of this project is to study how to integrate two approaches, namely belief compression and belief clustering, to address the intractability of POMDP problems. Techniques involved include data clustering, dimension reduction techniques and their integration into a value-directed framework for computing optimal policy for POMDPs.

Privacy Policy Enforcement in Data Analysis Workflows
	Staff	Dr. CHEUNG, William Kwok Wai (Supervisor of MPhil Student Mr. CHAN, Kai Kin)
	Abstract	To study a semantic approach for enforcing privacy policies in data analysis workflow systems. In particular, we introduce privacy preservation and analysis-relevant concepts as ontologies, incorporate them into a policy framework which represents polices, resolves potential conflicts among them, enforces them as well as provides suggestions on corrective actions.

Discrete integrable systems and their applications to convergence acceleration algorithms
	Staff	Dr. TAM, Hon Wah (Project Investigator)
	URL	http://www.comp.hkbu.edu.hk/~tam/rgc09-10_index.html
	Abstract	This research includes: (1) studying the integrable bilinear study of the Heisenberg-type equations and their discrete analogues. The emphasis is placed on integrable discretizations of some Heisenberg-type equations such as the Myrzakulov equations. (2) investigating some properties of discrete integrable systems that are related to the properties of convergence acceleration algorithms. We expect that discretizing integrable PDEs will lead to new sequence transformations.

On Motion Based Visual Feature Selections for Lip-Reading
	Staff	Prof. CHEUNG, Yiu Ming (Project Investigator)
	Abstract	Lip-reading has potential attractie applications in information security, speech recognition, and so forth. This project is to develop a new visual feature representation and extraction approach to the lip localization for lip-reading. The proposed method features high accuracy of lip localization and robust performance against the shadow. We build up a prototype of lip-reading system.

Associative Indexing for Semantic Visual Information Discovery
	Staff	Prof. LEUNG, Clement Ho Cheung (Project Investigator)
	URL	http://www.comp.hkbu.edu.hk/~asmimr
	Abstract	Current visual information retrieval is concerned with low-level features. The key to effective semantic visual information search critically depends on the indexing mechanism. In this project, we shall develop a methodology for supporting high-level semantic visual information discovery through the creation of associative indexes, which may be automatically built up and expanded by utilizing ontological data relationships and contextual knowledge.

Automatic Generation of High Precision Image Annotations through Concept Propagation
	Staff	Prof. LEUNG, Clement Ho Cheung (Project Investigator)
	URL	http://www.comp.hkbu.edu.hk/~asmimr
	Abstract	An effective mechanism of identifying image contents is semantic annotation. Without effective annotations, the semantic index cannot be built. The objective of this research is to address this problem by establishing algorithms for the automated generation of high-precision semantic image annotations. In doing so, high precision searching of a vast amount of otherwise unsearchable image collections would become possible.

Mining of Semantic Image Content Using Collective Web Intelligence
	Staff	Prof. LEUNG, Clement Ho Cheung (Project Investigator)
	URL	http://www.comp.hkbu.edu.hk/~asmimr
	Abstract	Human users spend a vast amount of time in interacting with image contents on the Web. Their interaction entails the exercise of considerable perceptive intelligence, visual judgment and mental evaluation. In this project, a collaborative indexing mechanism is developed whereby the aggregate intelligence and judgement of different Web users are continuously transferred to the Web.

Semantic Access and Characterisation of Web Images
	Staff	Prof. LEUNG, Clement Ho Cheung (Project Investigator)
	URL	http://www.comp.hkbu.edu.hk/~asmimr
	Abstract	The Web is becoming increasingly image centred. By injecting different levels of semantic characterisation into Web images by purely automatic means, this research will raise the value and utility of the Web, and to allow the discovery and deployment of an important portion of the Web which up to now has not been substantially explored.

Autonomy-Oriented Computing (AOC), Self-Organized Computability, and Complex Data Mining
	Staff	Prof. LIU, Jiming (Project Investigator)
	Abstract	Future challenges will lie in breakthroughs in scalable/robust computational solutions to complex data-mining problems that are: (1) Petabyte-scale (e.g., mining social networks), (2) dynamically-evolving (e.g., detecting social norms), (3) interaction-rich as well as trans-disciplinary (e.g., preventing economic/ecological crisis), and (4) highly-distributed (e.g., community evolution). This project concerns the development of an unconventional computing paradigm, called Autonomy-Oriented Computing (AOC).

Complex Emergent Behavior, Self-Organized Criticality, and Phase Transition in Multi-Agent Systems
	Staff	Prof. LIU, Jiming (Supervisor of MPhil Student Mr. HU, Bingcheng)
	Abstract	In this project, we address several issues on modeling complex or emergent behaviors in Multi-Agent Systems. In particular, we focus on the local-global relationship in a Self-Organizing Multi-Agent System (or SOMAS), a RoboNBA system, and study the phenomena of self-organized criticality (SOC) and phase transition in SOMAS.

Mechanism Design for Multi-Agent Systems in Distributed Data Mining
	Staff	Prof. LIU, Jiming (Project Investigator)
	Abstract	Methods for governing competitive systems have been studied in the field of distributed artificial intelligence (DAI), e.g., for autonomous agents and multi-agent systems. Mechanism design theories have been proposed and applied to networking and agent-based e-commerce. In this project, we are interested in investigating the use and impacts of mechanism design in distributed data-mining (DDM).