Integrated Devices for Artificial Intelligence and VLSI

8Advancement of Neuromorphic Computing Systems with Memristors

Jeetendra Singh¹, Shailendra Singh²*, Balwant Raj³, Vikas Patel¹ and Balwinder Raj⁴

¹Dept. of ECE, NIT Sikkim, Sikkim, India

²Dept. of ECE, Pranveer Singh Institute of Technology Kanpur, Kanpur, Uttar Pradesh, India

³Dept. of ECE, NIT Jalandhar, UIET, Panjab University SSG Regional Centre Hoshiarpur, Punjab, India

⁴Dept. of ECE, NIT Jalandhar, Punjab, India

Abstract

A computing device is deemed rudimentary if its operational scope is confined to elementary high-school-level mathematical tasks. Conversely, a mechanism that incorporates components of biotic origin, like as DNA units, as opposed to conventional electrical elements, closely mirrors the intricacies of the human brain. The instantiation of a computation system that operates neurons and synapses in parallel, exhibiting high capacity and low power consumption, constitutes a neuromorphic computing system (NCS). Neurons communicate through the transmission of both chemical and electrical signals. A neural network is established through minuscule junctions, termed “synapses,” where each neuron is intricately linked to others, facilitating signal propagation. Neuromorphic systems seamlessly integrate computing systems and memories, thereby mitigating the separation between them and overcoming the challenges associated with the “memory wall.” This chapter provides an exhaustive perspective on neuromorphic computing, commencing with an examination of recent challenges ...