MAX CRANSTON
News Writer

   You don’t think in black and white, so why should your computer?

Engineering professor Madan M. Gupta is a leader in work to make robots and computers think in a more complex way, by modeling it on human thinking. This year, he is being honoured as a distinguished chair for his over four decades of research in neuro-control systems and fuzzy logic.

His integration of a background in electrical engineering to neural systems, fuzzy logic and computing with words has established many pivotal concepts that are just beginning to be explored.

A professor emeritus and director of the Intelligent Systems Research Laboratory in the College of Engineering, Gupta has garnered international recognition for his research in the fields of fuzzy logic and neural networks.

“No matter how much I learn and publish, I feel as if I am just a baby in the field of fuzzy-neural systems. We have just barely scratched the surface of the field,” said Gupta.

Unlike traditional computing’s binary logic which has only two options — a one for yes or a zero for no — fuzzy logic uses graded membership between zero and one. This emulates the vagueness, or fuzziness, inherent in human language.

 Fuzzy logic represents a rational and logical decision-making system, with value somewhere between true and false — it takes into account the grey areas.

Take a light switch: turning the switch on and off is binary.  

The light switched on completes an electrical circuit and you get light — otherwise the circuit is incomplete, and no electricity flows.  

By contrast, if you were to use a dimmer switch, the value is no longer limited to on or off, but has different degrees of being turned on. It’s important to remember that fuzzy does not replace traditional logic, but provides more information about the current state. The light can still turn on or off, but there are different stages you can now reach within “on.”

 Fuzzy logic was first proposed in 1965 by professor Lofti A. Zadeh from the University of California. Gupta was introduced to the concept in 1968 when he first met Zadeh at the IFAC Congress held in Dubrovnik, in the former Yugoslavia. Zadeh is often referred to as “the father of fuzzy logic,” a term coined by Gupta for his friend and mentor.

The theory of neural networks reflects upon the mathematical understanding of the brain in decision making processes, whereas the mathematics of fuzzy logic provides some precision to the fuzziness in our thinking and cognitive processes, and thereby in our human language and the way we express our feelings.

“In Saskatoon, a temperature of -5 Celsius gives a feeling of ”˜very cold’ in the month of June, whereas this temperature is categorized as a ”˜fairly warm’ weather in the month of November,” said Gupta.

This concept is being applied to modern life in many ways. Car manufacturers have introduced fuzzy logic into braking systems. Anti-lock brakes use sensors which can examine specific parameters of fuzzy sets. Cameras, elevators and air-conditioners use fuzzy logic to operate.  

“In everyday life we use vague language and a machine such as the computer cannot understand this qualitative language, like ”˜today the weather is very warm,’ ” Gupta explained.

“The mathematics of fuzzy logic provides precision to the fuzzy statements.”  

Human language is extremely hard for a machine to process and understand. Gupta’s work and research are helping to create more intelligent system designs.  This is an emerging interest; it has been used commercially in text-to-voice software; though much of the technology is still in its infancy.

Spanning almost a half century, Gupta’s research has international significance. He served as a founding member of some international societies including the International Fuzzy Systems Association, North American Fuzzy Information Processing Society and Canadian Fuzzy Information and Neural Society.

– –
image: Danielle Siemens