Straying from the traditional von Neumann
computing architecture, IBM
researchers are now in the midst of developing cognitive computers, which
mimic processes of the human brain, through the use of neurosynaptic computing
Cognitive computing could open up a world of new
technologies in the future. These systems, which would not be programmed like
traditional computers, could learn through experiences like humans. They
could create hypotheses, find correlations, remember, and learn from its
environment while consuming less power and occupying less volume.
"Imagine traffic lights that can integrate
sights, sounds and smells and flag unsafe intersections before disaster happens
or imagine cognitive co-processors that turn servers, laptops, tablets and
phones into machines that can interact better with their environments,"
said Dharmendra Modha, project leader for IBM Research.
Now, IBM has brought these hypothetical scenarios
one step closer to reality with the creation of cognitive computer chips. These prototype
chips utilize digital silicon circuits inspired by neurobiology, but do not
contain biological elements. They contain a "neurosynaptic core,"
which consists of an integrated memory (mimicking synapses), communication
(mimicking axons) and computation (mimicking neurons). Through silicon
circuitry and advanced algorithms along with principles of neuroscience,
nanoscience and supercomputing,
these chips are capable of imitating biological processes such as those
occurring in the human brain.
So far, IBM has developed two working prototypes
that are currently undergoing testing. Both cores have 256 neurons and were
fabricated in 45 nm SOl-CMOS. One core has 262,144 programmable synapses while
the other contains 65,536 learning synapses. Researchers have already achieved
applications like machine vision, associative memory, classification,
navigation and pattern recognition.
"This is a major initiative to move beyond
the von Neumann paradigm that has been ruling computer architecture for more
than half a century," said Modha. "Future applications of computing
will increasingly demand functionality that is not efficiently delivered by the
traditional architecture. These chips are another significant step in the
evolution of computers from calculators to learning systems, signaling the
beginning of a new generation of computers and their applications in business,
science and government."
IBM's cognitive chips are part of the Systems of
Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE) project, which
aims to create a system that is capable of rewiring itself as it interacts
with its environment while still analyzing complex information from several
sensory modalities. In addition, the system must "rival the brain's
compact size and low power usage."
Currently, IBM has completed Phases 0 and 1 with
its cognitive chips for the SyNAPSE project. It was awarded $21 million in new
funding for Phase 2 by the Defense Advanced Research Projects Agency (DARPA).
For Phase 2, IBM will work with researchers from
Columbia University, Cornell University, the University of California - Merced,
and the University of Wisconsin - Madison. The team will work to achieve a
cognitive computing architecture consisting of an "on-chip network of
lightweight cores," resulting in one integrated system of software and
Eventually, IBM would like to develop a chip
system with ten billion neurons and one
hundred trillion synapses that occupies less than two liters of volume and
consumes one kilowatt of power.