ISc proposes human brain inspired computing platform to boost AI tool

Contact Counsellor

Last Updated16/09/2024

Researchers from the Indian Institute of Science (IISc) have developed an analogous computing platform within a molecular film which is able to mimic a human brain-like functioning.

In a groundbreaking advancement in Artificial Intelligence (AI), researchers at the Indian Institute of Science (IISc) have developed an analogous computing platform within a molecular film capable of mimicking the functioning of the human brain.
This innovative technology, designed at the Centre for Nano Science and Engineering (CeNSE) in Bengaluru, marks a major breakthrough in neuromorphic computing, a field that aims to replicate human brain-like processes in electronic systems.

The research team has successfully created a ‘brain on a chip’ design, which although analog, performs tasks such as processing and data storage in ways that closely resemble the human brain.
The platform stands out with its ability to offer 16,500 conductance states in contrast to the binary 0 and 1 states used in conventional digital computing systems. These digital systems, while powerful, often consume substantial energy and time, leading to slower processing speeds compared to the anticipated capabilities of this new AI-based platform.

This advancement has the potential to transform AI applications by enhancing flexibility and deployability across personal electronic devices, such as smartphones, laptops, and desktops.
According to the researchers, the platform could have a significant impact on machine learning and scientific computing, addressing key challenges faced in current neuromorphic systems.
The primary advantage of this technology lies in its efficient use of molecular films, which enable researchers to trace free ionic movements, creating an extensive network of memory pathways.
This development overcomes the energy-intensive nature of existing digital platforms, making it a more sustainable and scalable option for future AI tasks.

One of the key aspects of this breakthrough is the precise control the team achieved over molecular transitions within the film. Using timed voltage pulses, they were able to map a wide range of molecular movements to distinct electrical signals, essentially creating a "molecular diary" of states.
These intermediary states, which were previously inaccessible, represent a significant leap beyond the binary limitations of digital computing.
Sreebrata Goswami, from CeNSE, explained that by controlling molecular kinetics within a circuit powered by nanosecond voltage pulses, the team was able to generate an unprecedented number of unique memory states.
This allowed them to create a highly efficient neuromorphic accelerator that stores and processes data in the same location, much like the human brain.

The neuromorphic accelerator developed by IISc researchers is expected to integrate seamlessly with existing silicon circuits, significantly boosting their performance and energy efficiency.
In the realm of AI, this innovation could unlock new milestones in neuromorphic computing, including advancements in edge training, long-short-term memory models, generative adversarial networks (GANs), and transformer models.
This indigenous effort addresses several challenges faced in AI and computing and positions India at the forefront of cutting-edge technological progress in neuromorphic computing.