Indian Institute of Technology Delhi, New Delhi
Debanjan Bhowmik is currently an Assistant Professor in Department of Electrical Engineering, Indian Institute of Technology–Delhi, working in the area of spintronics-based low-power computing. He supervises research in the area of spintronics-based hardware neural networks. He obtained his B. Tech. in Electrical Engineering from IIT–Kharagpur in 2010 and PhD from University of California–Berkeley in 2015, working in the field of nanomagnetism and spintronics. He was elected as an Associate of the Indian Academy of Sciences in 2017
SESSION 1C: Inaugural lectures by Associates
V C Thakur, Dehradun
Hardware implementation of neural network algorithms, based on machine learning and neuroscience models, using spintronic devices as non-volatile memory elements
Neural Network algorithms are widely used by the machine learning and data sciences community to solve various classification, recognition and prediction tasks. These algorithms, if implemented on hardware instead of software, can provide further advantages owing to the parallel architecture and the principle of memory embedded computing inherent in these algorithms. Spintronic devices owing to their non-volatility can make excellent memory elements for the hardware implementation of Neural Networks. The speaker will discuss a few such spintronic implementations of Neural Network, which he and his team have carried out through simulations at IIT Delhi. One such implementation follows a Neural Network algorithm that is largely inspired by the functioning of the brain – Spiking Neural Network, enabled with Spike Time Dependent Plasticity. If implemented in spintronic hardware it can not only solve several problems of relevance to the machine learning community but also may help improve our understanding of the functioning of the brain in future.