Spiking Neuron for Fast, Accurate AI


The human brain is one of the most powerful and intelligent natural computers known to mankind. Neuromorphic Computing refers to the field of technology in which engineers try to build intelligent machines that are inspired by the work of mammalian brains. Neurons and synapses are considered to be the most important building blocks that generate intelligence in the brain.

Researchers at IIT Delhi under the direction of Prof. Manan Suri, Faculty of Electrical Engineering, have invented a new spiking neuron model called DEXAT (Double EXponential Adaptive Threshold). The invention is important as it helps build accurate, fast, and energy efficient neuromorphic artificial intelligence (AI) systems for real world applications such as speech recognition.

The interdisciplinary work lies at the interface between AI, neuromorphic hardware and nanoelectronics.

“We have successfully demonstrated that storage technology goes beyond simple storage. We have used semiconductor memory efficiently for applications such as in-memory computing, neuromorphic computing, edge AI, sensor technology, and hardware security. This work specifically uses analog properties of nanoscale oxide-based memory modules to build adaptive spiking neurons, ”says Suri in a press release from IIT Delhi.

The study demonstrated a neuron model with higher accuracy, faster convergence, and flexibility in hardware implementation when compared to other modern threshold adaptive spiking neurons. The proposed solution achieves high performance with fewer neurons. The advantages of the proposed invention have been demonstrated using several data sets.

The scientists successfully demonstrated a hybrid hardware implementation based on nano devices. It was found that the proposed neuromorphic network for nanodevices achieves an accuracy of 94 percent even with very high device variability, which indicates robustness.


Leave A Reply