Tutorial AM4

Low-power Adaptive Neuromorphic Systems: Device, Circuits, Algorithms and Architectures

Arindam Basu, Nanyang Technological University

Abstract: The recent success of “Deep neural networks” (DNN) has renewed interest in bio-inspired or neuromorphic machine learning algorithms. However, most work has focused on designing energy efficient circuits for inference while the learning is relegated to a server working on offline collected data. This method has drawbacks in terms of scalability of IoT as well as data non-stationarity in biomedical devices (both wearable or implantable). Hence, this tutorial describes in detail strategies to design adaptive neuromorphic systems that learn from data in an online, unsupervised or semi-supervised fashion. To reduce power dissipation and hardware resource usage, cross-layer innovations spanning novel devices (e.g. floating-gate, memristor etc), circuits (switched capacitor synapse, transposable SRAM synapse, bi-stable synapse, neuron and driver circuits etc), algorithms (Spike time dependent plasticity, Spike Time and Rate based plasticity, Least mean square, Backpropagation and variants like feedback alignment etc) and architectures (crossbar, island style etc) will be discussed. Finally, we will present some case studies of usage of adaptive neuromorphic systems in brain-machine interfaces, robotics and the internet of things.


Arindam Basu

He has been working in the area of neuromorphic systems for more than a decade and his work has been recognized by several awards including MIT Technology Reviews TR35@Singapore award in 2012. He has also delivered tutorials at IEEE BioCAS 2010, IEEE ISBB 2014 and IEEE IJCNN 2015 on related topics of neuro-inspired circuits. He has been serving as IEEE Distinguished Lecturer for Circuits and Systems during the term of 2016-17 and has recently guest edited (corresponding guest editor) a special issue on the topic of Adaptive Neuromorphic Systems in IEEE Journal on Emerging Topics in Circuits and Systems. He was also a topic leader in the NSF sponsored Telluride Neuromorphic Engineering workshop in 2015.

Arindam Basu received the B.Tech and M.Tech degrees in Electronics and Electrical Communication Engineering from the Indian Institute of Technology, Kharagpur in 2005, the M.S. degree in Mathematics and PhD. degree in Electrical Engineering from the Georgia Institute of Technology, Atlanta in 2009 and 2010 respectively. Dr. Basu received the Prime Minister of India Gold Medal in 2005 from I.I.T Kharagpur. He joined Nanyang Technological University as an Assistant professor in June 2010 and is currently a tenured Associate Professor at the same university.

Dr. Basu received the best student paper award at Ultrasonics symposium, 2006, best live demonstration at ISCAS 2010 and a finalist position in the best student paper contest at ISCAS 2008. His research interests include bio-inspired neuromorphic circuits, non-linear dynamics in neural systems, low power analog IC design and programmable circuits and devices. He is currently serving as an Associate Editor for IEEE Sensors journal, IEEE Transactions on Biomedical Circuits and Systems and Frontiers in Neuromorphic Engineering. He is a member of the IEEE CAS Technical Committees of Sensory Systems, Biomedical Circuits and Systems and Secretary for Neural Systems and Applications. He has served as Technical Program Co-chair for IEEE International Symposium on Integrated Circuits (ISIC) 2016, Publicity-co-chair for IEEE ISCAS 2017 and is serving as Special Sessions co-Chair for IEEE BioCAS 2018. He has been guest editor for two special issues in IEEE Transactions on Biomedical Circuits and Systems for selected papers from BioCAS 2015 and ISCAS 2015.