Search

Brain-inspired computing is attracting considerable attention because of its potential to solve a wide variety of data-intensive problems that are difficult for even state-of-the-art supercomputers to tackle. The ability of the human brain to process visual and audio inputs in real time and make complex logical decisions by consuming a mere 20 W makes it the most power-efficient computational engine known to man. While state-of-the-art digital complimentary metal-oxide-semiconductor (CMOS) technology permits the realization of individual devices and circuits that mimic the dynamics of neurons and synapses in the brain, emulating the immense parallelism and event-driven computational architecture in systems with comparable complexity and power budget as the brain, and in real time, remains a formidable challenge.

While machine learning algorithms are being successfully employed for a wide variety of high level cognitive tasks, their wide-spread adoption on mobile platforms faces challenges because of the large network sizes and extremely long training times involved in learning underlying features of user-dependent data. In this paper, we first discuss the requirements and target specifications of nanoscale devices that could enable hardware platforms supporting on-chip, real-time learning based on third generation spiking neural networks. We then survey some recent developments from the field of emerging non-volatile memory technologies such as chalcogenide based phase change memory and other memristive devices that could be used to build cognitive learning systems. We will then discuss how we have implemented synaptic plasticity in a Cu/SiO2/W memristive device at voltages below 500 mV using bio-mimetic waveforms. These devices hold promise to herald the next generation of information processing systems that are inspired by the brain.

How do we design cognitive computing systems that not only deliver well-calibrated and trustworthy decisions, but can also be deployed in the field, are capable of adapting or learning the features of unseen input streams in a secure manner, and operate within the energy budgets of battery-operated devices?

In this paper, we review some of the novel emerging memory technologies and how they can enable energy-efficient implementation of large neuromorphic computing systems. We will highlight some of the key aspects of biological computation that are being mimicked in these novel nanoscale devices, and discuss various strategies employed to implement them efficiently. Though large scale learning systems have not been implemented using these devices yet, we will discuss the ideal specifications and metrics to be satisfied by these devices based on theoretical estimations and simulations. We also outline the emerging trends and challenges in the path towards successful implementations of large learning systems that could be ubiquitously deployed for a wide variety of cognitive computing tasks.

While machine learning algorithms are being successfully employed for a wide variety of high level cognitive tasks, their wide-spread adoption on mobile platforms faces challenges because of the large network sizes and extremely long training times involved in learning underlying features of user-dependent data. In this paper, we first discuss the requirements and target specifications of nanoscale devices that could enable hardware platforms supporting on-chip, real-time learning based on third generation spiking neural networks. We then survey some recent developments from the field of emerging non-volatile memory technologies such as chalcogenide based phase change memory and other memristive devices that could be used to build cognitive learning systems. We will then discuss how we have implemented synaptic plasticity in a Cu/SiO2/W memristive device at voltages below 500 mV using bio-mimetic waveforms. These devices hold promise to herald the next generation of information processing systems that are inspired by the brain.