Dual Goals:

- Understanding brain information processing mechanism
- Developing brain-like intelligent systems (Artificial Brain and Artificial Cognitive Systems)

Research Areas:

   The main research areas reside in computational models of brain information processing mechanism and their applications to build human-like intelligent systems, i.e., Artificial Brain and Artificial Cognitive Systems (ACS). These functional models are based on information theory and inspired by findings in cognitive science. Intelligent robots with human-like cognitive functions are examples of ACSs, which improve their functional ability by learning from users and other ACSs.
   Although human-like perception has been regarded as the main achievements on the laboratory, recently the research topics extends further into the higher brain functions including knowledge, emotion, consciousness, and human behavior. In collaboration with American-based NeuroSky Inc., it also works on brain-machine interfaces and neurofeedback mind self-controls both based on EEG and possibly eye-movement.

Main Achievements:

- auditory models for speech feature extraction, sound localization and blind signal separation
- top-down selective attention model for robust recognition. (How people see what he/she wants to see/hear?)
- multi-modal fusion based on the top-down attention (i.e., audio-visual integration for lip-reading)
- feature extraction, selection and adaptation for image, text, emotional speeches, music, and EEG.
- neuromorphic chips and boards based on the developed models
- ABrain (Artificial Brain) and OfficeMate (Artificial Secretary) as a testbed of human-like intelligent systems

On-Going Research Projects:

(1) Artificial Cognitive Systems: Proactive Learning and Situation Awareness for Robots (2009-2012)

- feature extraction for the recognition of secondary phenomena (such as emotion in speeches and EEG signals,
  facial expressions, and musical timbers, etc.)
- proactive learning algorithm with internal state models for self-identity, emotion, and motivation
- knowledge representation and incremental self-learning
- decision-making and socialization models (among people and between human and robots)
- detection of intentional and un-intentional human desires for intelligent user interface (based on EEG,
  eye-movement, etc.)
 

(2) Blind Signal Separation and Noise Canceling in Home Environment (2009-2013)

- signal separation and enhancement based on Independent Component Analysis (ICA) with additional constraints
  and/or top-down attention (mainly for speech and EEG signals)
 

(3) EEG-based Brain-Machine Interface with NeuroSky Headset (2008-?)

- applications of dry-electrode EEG headset for human-oriented user interface for
  game/toys and neurofeedback mind controls (in collaboration with NeuroSky Inc., USA)