CHENNAI: Google DeepMind’s AlphaGo, an artificial intelligence programme developed using deep neural networks and machine learning techniques, hit global headlines last year when it beat South Korean Go grandmaster Lee Sedol to win the series 4-1.
However, not many know that AlphaGo has consumed a whopping 30,000 watts of power to complete the task, while the human brain consumes around 20 watts!
What gives the human brain such efficiency has so far proven elusive to replicate in computers. Not surprisingly, man’s most defining organ is also the least understood. Although an adult human brain weighing 1.4 kg is made up of close to 100 billion neurons, scientists do not know how many different kinds of human neurons exist.
The Neuroscience Information Framework, a web-based inventory of the National Institutes of Health (NIH) of the US Department of Health and Human Services, has so far been able to identify just 800 types of neurons.
In India, which is slowly gaining pace in the emerging field of computational brain research, a group of about 25 people, including students and faculty members, at the Indian Institute of Technology-Madras (IIT-M) is chasing the dream of developing brain-inspired innovations. The Centre for Computational Brain Research (CCBR) was set up in 2015 with generous funding from co-founder of Infosys Kris Gopala-krishnan. Three Chairs each with an endowment of `10 crore were constituted.
Interestingly, all three Chairs headed by India-born American scientists coming from different disciplines, are actively trying to establish a two-way interface between the fields of neuroscience and engineering.
In an interview with Express on the sidelines of the 2nd annual workshop on computational brain research at IIT-M, Anand Raghunathan, Professor of Electrical and Computer Engineering at Purdue University in Indiana and heading the C R Muthukumar Distinguished Chair in the CCBR, said there was a need to make energy efficient radical computer systems because of explosion of big data. Or it will not sustainable.
“If you see the projected rate of growth in computational requirements for deep neutral networks for next five years, the amount of energy likely to be consumed is unsustainable. Particularly, this comes at a time when progress in semi-conductor technology is slowing down.
The Moore’s law, which has made computing technology and industry what it is today, is unlikely to hold good for long,” he said and added that understanding of human brain’s structure and neural circuitry was the key to improving computer speed and energy efficiency and developing next-generation algorithms and software and hardware for machine intelligence.
However, this is not a easy task to achieve. Unravelling mysteries of human brain is probably the biggest challenge of this century. Mriganka Sur, professor at MIT, Cambridge, and also the head of the NR Narayanamurthy Distinguished Chair at IIT-M, told Express that the study of brain has a long history, at least 100 years. Spanish neuroscientist Santiago Ramon y Cajal was the first person to have been able to mark the brain cells and describe beautifully the microscopic structure of the brain comprising 100 billion neurons for which he was awarded Nobel Prize in 1906.
“Each cell is a separate entity. It processes information. It speaks to thousands of other cells and hundreds of synapses or connections. A brain has 1,000 trillion synapses and takes a lot of processing, which is the reason why our brain does what it does and there are certain common principles that drives the entire activity. The goal of computational brain research is to discover these principles of information.
The fact that we see, hear and talk is because light is converted into electricity in the eye and the electricity goes into the brain and all that we see is encoded in the activity that eye provides the brain. There are several things that we can’t see because eye don’t encode them like wavelengths of infrared, but other birds see them and there are things that we don’t hear, but dogs do. These are deep mathematical and computational principles that we can use to understand these processing and if we can understand how the brain works we can produce better machines that can utilise some of these principles and perhaps treat brain disorders,” Sur explained.
Kris Gopalakrishnan, during a talk at the workshop, said the three Chairs in computational brain research at IIT-M provided perfect platform for neuromorphic computing, where physicists, electrical engineers, scientists and biologists are involved.
Mriganka Sur and another Chair headed by Partha Mitra of Cold Springs Harbor Laboratory in New York are also part of BRAIN initiative, launched by former President Barack Obama in 2013, and some of the data generated in US was brought to IIT-M for further analysis.