Neuromorphic computing with memristor devices

Neuromorphic computing is an approach to artificial intelligence (AI) that uses hardware elements inspired by the components of the brain and presents an alternative to the dominant von Neumann computing paradigm based on digital hardware. Recently, nano-scale devices known as memristors have been identified to offer area- and energy-efficient solutions when used in neuromorphic computing circuits that would require circuitry consuming hundreds of µm2 of die area to replicate with digital CMOS. Measurement of the static and dynamic electrical properties of memristor devices is therefore a strong concern for developing neuromorphic capabilities. However, given the stochastic nature of memristor switching, acquisition of massive amounts of switching data and a statistical treatment of that data is required for characterizing the devices and designing circuitry based on them. Currently, the lack of efficient testing and data acquisition methods is a major bottleneck facing the development of memristor-based neuromorphic devices.

Faculty Supervisor:

Dominique Drouin

Student:

Partner:

Universidade Federal de Minas Gerais

Discipline:

Engineering

Sector:

Education

University:

Université de Sherbrooke

Program: