Fuzzy logic: Prof. Sankar Pal

Adaptive. Intelligent. Quick learner. If you think all these attributes are only unique to humans, you are mistaken. Mimicking the imprecise decision-making process of the human brain, scientists have devised new ways that coax computers to think beyond the confines of the digital world. “Collectively known as soft computing, these computing tools empower a digital machine to deal with imprecise and vague data that abound in real life,” said Dr M.G.K. Menon, president of the Indian Statistical Institute.

He was speaking at a seminar held recently as part of the inaugural programme of the ‘Centre for Soft Computing Research: A National Facility’ at the ISI, Calcutta.

“Soft computing is still very much a software-based technology,” said Dr Sankar Pal, director, ISI, while addressing the seminar. “It relies on the existing hardware of digital computers, but increases machines’ intelligence to a great extent.”

Pal explained that the main components of soft computing are fuzzy logic, genetic algorithm, artificial neural networks (ANN) and probabilistic reasoning. “The new soft computing centre at the ISI here will open up fresh avenues of research in bioinformatics,” he said.

According to him, soft computing tools like genetic algorithms, designed to simulate processes in natural environments that favour the survival of the fittest animal, will help us assemble a specific group of genes from the pool of 30,000 genes that reside in every human cell ? for instance, the genes that make the blood proteins or the ones that are active in the liver. “Being an effective searching tool, genetic algorithm hits upon the target solution from myriad options for a given problem,” Pal said.

Besides GA, fuzzy logic is another important tool for soft computing. “Unlike digital reasoning, which requires ‘yes’ and ‘no’ values (0s and 1s, or, more specifically, on-off electrical circuits), fuzzy logic can handle ambiguous values like ‘maybe’, ‘nearly’, and ‘very much’ (every possible value between 0 and 1),” Pal commented.

Fuzzy logic often uses if-then rules. For example, consider the case of the temperature regulator that uses a fan. If it is very cold, the fan stops running. If it is hot, it the fan speeds up. Fuzzy logic chips are used in colour televisions, refrigerators and air-conditioners.

“But fuzzy logic has a drawback. It can’t learn,” Pal pointed out. “This obstacle is overcome by merging fuzzy logic with ANN, a real electronic circuit made of physical nodes (chips), analogues of neurons and their connections in human brains.”

According to him, this gives rise to what is known as a neuro-fuzzy system, which is being used to replace fuzzy logic-based applications in electronic gadgets like colour televisions and vacuum cleaners.

Biplab Das