Harnessing the power of mind-mimicking computers

Combining elements of biological models and the processing power of computer systems opens a broad realm of possibilities that allow the incorporation of learning and decision-making capabilities into a machine.

Nadipuram Prasad, associate professor of electrical and computer engineering and master’s candidates Salvador Almanza and Sai Nikhil Kuchi are working on the development of what is known as MIND systems (Machine INtelligence for Decision-making), use of soft-computing systems that can solve highly complex problems through the emulation of biological models. Applications of such systems range from home appliances and autonomous vehicles to smart robots and threat detection devices.

Soft-computing is the ensemble of fuzzy logic, neural networks and evolutionary computation (genetic algorithms), each representing a type of biological mechanism, that can be incorporated in computer-based systems. These bio-inspired models can be easily implemented by software and are not limited just to high-performance computers, but also to embedded systems that incorporate smaller processors or microcontrollers in their architectures.

“All decision-making is based on fact and rules,” explained Prasad. “For example, humans make decisions in the form of ‘if-then’ rules and have the advantage that data can be extracted by directly asking the expert about how control actions are executed. In other biological systems, information about the behavior is available only through observation, making the system design more challenging. Humans and animals also have the ability to recall past events and make decisions based on information drawn from past events. This pattern recognition or recalled learning can be mimicked by neural networks, which can extract data, recognize patterns and process information.”