The story of Fuzzy Logic is interesting. Prof. Lotfi Zadeh developed a new logic and set theory to address multi-valued variables and called it Fuzzy Logic. Westerners termed it permissive as they are constrained in their thinking in a binary mode whereas easterners didn’t have any such problem in accepting the new theory.
Please see below my Key Note on this topic published in Computers Today, October 1995.
Article : How Fuzzy is this Logic?
During my doctoral studies at Oklahoma State University in the US, in the early 1970's, one area of research which excited me was Fuzzy Logic. Although its usage at this point of time was not quite widespread, this does not hold true anymore. Today even in India, advertisements of consumer durables such as air-conditioners, washing machines and rice cookers claim that the products use “Fuzzy Logic” or even “Neuro Fuzzy Logic”. What is the Fuzzy Logic all about? And why has it taken so long to exploit?
Not So Fuzzy
The binary logic, or set theory on which modern computers are based, professes that all things can be classified into two categories — true or false. For example, the colour of my shirt is white (true) or non-white, say black (false). An electrical circuit signal voltage is high or low. However in real life, everything is a matter of “degree”. So, the colour of my shirt could be neither white nor black, but grey. The signal in the electrical circuit is somewhere between high and low. It is not possible to define these ‘grey’ facts in real world with the ‘black and white’ or binary science.
Professor Lotfi Zadeh of University of California at Berkeley developed a new logic and set theory to address these multi valued situation, and called them Fuzzy Logic or Fuzzy Sets. In this scheme of things, conventional logical binary and set theory become one special case. However, this theory received brickbats from western scientists for triggering off the scientific revolution. There were no Government or Industry grants for future research. Scientific journals and conferences refused to accept papers on Fuzzy Logic branding it “permissive”.
Later Professor Bart Kosko, faculty member of the University of Southern California, wrote a book “Fuzzy Thinking” in which he dispelled the aura surrounding Fuzzy Logic. He argued that today’s science suffers from a binary reflex — the world is either black or white, right or wrong, all or nothing. In contrast, eastern philosophy as articulated by Buddha sees the world in a different perspective — as a unit that embrace yin and yang at the same time. It sanctions contradictions, endorses ambiguity and demands that we get comfortable with our uncertainties. He concludes that eastern countries like Japan and Korea embraced Fuzzy Logic wholeheartedly because there was no disconnect between philosophy and science.
Fuzzy Logic lets computers reason with vague concepts like ‘somewhat cool’ or ‘very slow’. In traditional system, we use the set-point control scheme where binary logic is used to see whether the desired state is above or below the set point. Take the example of room air conditioner we may choose the set-point to be at ‘high or medium or low’ once the thermistor sensor gives a single indicating that the room temperature has gone above the set-point, the condenser gets activated. Similarly, when the temperature goes below the given limit, condenser is shut off. Thus, control actions are transient (on/off) in nature and induce stress on the system leading to reduction of its life span. Moreover, what we actually need is a ‘comfortable temperature’ and not a band of temperature, say between 15 and 25 degree centigrade. What is comfortable in day time may not be comfortable at night. Similarly, it can vary in winter as compared to summer. It also can vary from percent to percent. In other words, we want the set-point itself to be variable rather than fixed. Such controlled are called non-linear and conventional logic is unable to do a good job of this.
Neutral network-based technologies allows continuous monitoring of
temperature and application of appropriate cooling machanism resulting in a
smoother control. This is made possible via an intelligent algorithm stored in a microcontroller. A neural network such as this mimics the human brain and learns Fuzzy rules from experience, thus raising the IQ of the machine. In addition, it achieves the result in a fast, accurate and intelligent manner at performance levels previously unattainable. Moreover, with the present-day semiconductor technology implementation of neural network with Fuzzy Logic cost very little in compared to immense benefits one can derive.
Some Japanese and Koreans corporation have already reaped billions of dollars from designing washing machines that adapt to the load, television sets that tune itself against variation in signal strengths of different channels and camcorders that auto-focus and stabilize themselves on their own. Smart machines like these are going to be a part of next wave of consumer and communication technology bringing in an era of commercial machine intelligence.
There is also a great opportunity for those Indian companies which can quickly build expertise in Fuzzy Logic and Neural networks. We can work with other Asian neighbours like Japan, Singapore, Taiwan and Korea, to add ‘intelligence’ based on Fuzzy Logic to inexpensive and quality hardware being designed and manufactured there for the world. Any takers?
Dr. Sridhar Mitta is the Managing Director and Founder of NextWealth Entrepreneurs Private Limited. NextWealth provides Digital Business Process Management solutions by bringing in Human Touch to Digital Processes. He was the Chief Technology Officer of Wipro since its inception in 1980 for over 20 years.