Theory: Understanding the Fundamental Mechanisms of Brain-like Intelligence

Development of intelligent systems and discovering mechanisms for intelligent behavior is one of the most exciting research areas in science and engineering. With the recent development of brain research and modern technologies, scientists and engineers will hopefully find efficient ways to build brain-like complex systems that are highly robust, adaptive, and fault tolerant to uncertain environments. However, although scientists and engineers have successfully borrowed some ideas from biological intelligent systems, for instance, the designing of the insect-inspired robots, there is still no clear picture about how to design the brain-like intelligent machines. The biggest challenge comes from how to develop the models, algorithms, architectures, and organizations that are able to learn information, accumulate experiences, and make associations and predictions to accomplish desired goals (learning-memory-prediction framework), which are the critical elements for any biological intelligent systems.

To address the fundamental issues of understanding brain intelligence and finally toward building brain-like systems, we are particularly interested in: 

• Dynamic, incremental, and adaptive learning.

• Bio-inspired learning mechanisms;

• Hierarchical organization for learning, memory and prediction;

• Embodied intelligence;

• Value systems and goal-driven learning;

• Self-organizing associative memory architecture;

Design: Advanced Systems Prototyping, Design and Testing

Intelligent system models can be simulated in a sequential computer, mapped into a programmable hardware, or built in hardware. Software implementation is the easiest one but it has its inherent limitations. Today's software simulation can only handle small size of networks needed to implement brain-level intelligent systems. With the development of reconfigurable FPGA technology and VLSI technology, it is technologically possible to design brain-level complexity systems ("silicon brain") using such hardware technology in the future. .

To address the critical design issues, we are particularly interested in:

• Hardware-oriented intelligent architectures;

• FPGA based intelligent systems prototyping and testing;

• VLSI design of intelligent modules;

• Software/hardware co-design and simulation;

Applications:

Computational intelligence have wide applications including biomedical engineering, financial engineering, security systems, decision making process, etc.  We are currently explore different applications in the following areas:

• Biomedical data analysis

• Pattern recognition and classification;

• Decision making with information ambiguity;

• Financial data mining;

• Mobile sensor networks;

Publications

A list of recent publications can be accessed here.

Facilities

Lab facilities at SAIS can be accessed here.