19.03.21 - Brain-inspired computing with resistive memory

Irem Boybat (IBM Research Europe)

AI systems managed to reach and even exceed human performance in
various cognitive tasks, ranging from image recognition to strategic games
and to reasoning. Yet, today’s computing systems based on the classical
von-Neumann architecture dating from the 1940s cannot efficiently address
these highly data-intensive workloads. It is becoming increasingly clear
that we need to transition to non-von Neumann architectures in which
memory and logic co-exist in some form. Brain-inspired computing is one
such approach, where inspiration is taken from biological observations of
the brain. One can build in-memory computing units where the separation
between memory and processing is blurred, and physical attributes and
state dynamics of memory devices are exploited to perform certain
computational tasks. The neural structure and operation of the brain
including the rich neural and synaptic dynamics can also be adopted to add
to the information processing abilities and improve the efficiency of
computing systems. Resistive memory is expected to play a key role for
brain-inspired computing, from building in-memory computing arrays to
emulating neurons and synapses for neuromorphic computing. Explorations in
device technology and memory architectures could further enhance the
capabilities of brain-inspired computing systems.