Brief History: About ME

I had the opportunity of an enriching education that spanned across the world. I obtained a BSc in Computer Technology from the US, a Master of Engineering in EEE from Singapore and a PhD in Computing from Loughborough University. Immediately after my graduation from Loughborough in 1995, I worked for Gintic Institute of Manufacturing Technology till 2001. Gintic has since been renamed to SIMTech, a research institute under the A*STAR SIngapore. I subsequently joined the School of Computer Engineering, NTU, as an Assistant Professor. I began to head the Programme for Cognitive and Neural Systems within the Emerging Research Labs at NTU in 2005. Over the time, the programme continued to mature with an eventual funding of more than S$4 million worth of projects. I have since been given the opportunity to be the Director of the Emerging Research Labs in mid 2007. I was also previleged to win the prestigous Nanyang Award for Excellence in Teaching, 2006

I have had 7 years of working closely with the industry from 1995 to 2001 and have continued to maintain close contacts even after joining NTU in 2001. During my work with the industry, my projects included working with clients in the back-end silicon industry and also with others such as Segate, HP, IBM. Most of my projects involved simulation, optimzation and applications of neural networks. Towards the latter part of 2001, I was involved with algorithmic path planning for a 10 ton autonomous vehicle. This provided me the hands-on experience in real-time critical systems. In 2007, I was directly involved in bringing about S$700K worth of funding to design cognitive autonomous systems for real world applications.

The principles that drive me forward are essentially practical solutions to real-world problems. The exposure to the industry has given me ample scores of real-world problems. The desire to seek viable solutions thus generates the pursuit of methods and these often surface as my research objectives and goals. I am very practical oriented with all my research interests eventually ending in a real-world implementation.

Current Interests: Cognitive Research and Applications.

I had 4 enjoyable years of working on a full scale tracked vehicle that was capable of carrying up to 10 people. It was designed to navigate and traverse through tropically lush terrains to up to 1 km autonmously at speeds of up to 15 km/h. The industrial project was successfully closed in 2004. This interesting project left me with several unanswered questions that made me feel that much more was needed in the field of computer vision and neural network intellgence. Perhaps it was in the domain of Human Interface Technology. Perhaps we could even be more specific: The Cognitive Intelligence.

I have been seeking answers in the areas of biological vision and hippocampal inspired processes since 2005 when I was given the opportunity to head the Cognitive and Neural Systems Programme. It was also an extreme previlege to work with Professor Stephen Grossberg and Professor Gail Carpenter from Boston University's Department of Cognitive and Neural Systems. The investigations have since provided us some leads that allow our in-house systems to function successfully. This quest will continue with the objective of building up an understanding of how perceptive interpretations are derived by the time the neural signals arrive at the prefrontal cortex. I have decided to focus in areas of Cognitive Visual Understanding and Cognitive Navigation. I hope the videos in each of these areas will help you appreciate the significance of these developments. Due to the paradigm shifts in our approach, we show that these feats are able to address issues that were once impossible to resolve generically using conventional engineering methods.

The Cognitive Visual Understanding track has taken a dive into the deep end where we have just discovered the possibility of binocular image correspondence without the need for registering objects on either cameras. Unlike systems that use epipolar geometry and optical calibration, vergence in CogV can be performed in the log-polar transformed Visual Cortex Model. This model however is further enhanced using a biologically inspired receptive field implementation which exhibits robustness and accuracy.

The Cognitive Navigation track has also taken a twist from the conventional SLAM methods where active filtering of sensor probabilities are used to keep track of localization parameters. CogNav uses the principle of neuron-based place cells and coupling that with our proprietry K-Iterations Neural Network, we are able to resolve even the kidnapped robot problem.

A third component that we have recently began in our quest for cognitive intelligence is the study of attention. It has been long discovered that attention is the means by which we are alerted to an environmental stimulus. Through such a stimulus, the system learns to either respond, or ignore the event. Such attentive cues may come in the form of situational failures or successes, which trigger a learning process. By building attention into physical systems, we hope to examine how architectures that are built using cognitively enabled components, such as CogV or CogNav, will learn to derive context and engage in goal oriented behaviour. The difference now, is that the system determines for itself what it wants to accomplish, rather than what it was hard-coded to do.

Current Neural Network Research and Applications

We are still actively working on the K-Iterations Fast Learning Artificial Neural Network (KFLANN) and have been previleged to work closely with Professor Jacek Zurada to publish it. We are still undergoing investigations on the KFLANN and are about to release a morepotent version of the network that provides a means for tunning the classifier accuracy (categroy handling).

Work is currently being done to model the KFLANN as a Short-Term Memory (STM) module that interfaces the Long-Term Memory (LTM) Backpropagation (BP) model.

Cognitive Autonomous Rescue Vehicle

I was fortunate to be able to lead a 25-man Tech-X Challenge team, comprising of professors and students. Along with us were some industrial partners from ZionTECH Pte Ltd and EADS Singapore. The total funding for the project was equivalent to S$400K. We designed and built the actual vehicle from mere concepts and it is now able to navigate autonomously using its internal navigation algorithms. I have attached some photos on this website, but I suspect you may want to see and know more. The TechX Website is attached.

Site Last Updated June 2008