An IBM computer modeled after the human nervous system has been sold to the Lawrence Livermore National Laboratories (LLNL) in order to work on nuclear deterrence and digital security issues around nuclear weapons.
This is the first time that IBM has sold an offshoot of its IBM Research work to another entity. Specifically, IBM Research’s TrueNorth chip will be used in conjunction with the National Nuclear Security Administration to explore cybersecurity and to prevent the increase of nuclear weapons. The chip is “neurosynaptic,” which means that it is capable of “brain-like” pattern detection and sensory recognition. The TrueNorth system that LLNL will receive can process at speeds comparable to 16 million neurons and 4 billion synapses and takes just 2.5 watts of power.
Within that digital network in a single TrueNorth processor are 1 million “digital neurons” made up of 5.4 billion transistors. Like the human brain, they communicate with one another using electrical signals – in TrueNorth’s case, there are 256 million electrical signals running at 46 giga synaptic operations per second. It takes less energy than neural networks run on conventional computers, consuming 70 milliwatts.
“Neuromorphic computing opens very exciting new possibilities and is consistent with what we see as the future of the high performance computing and simulation at the heart of our national security missions,” said Jim Brase, deputy associate director for Data Science at LLNL. “The potential capabilities neuromorphic computing represents and the machine intelligence that these will enable will change how we do science.”
LLNL and IBM have worked together before: the TrueNorth processor was originally simulated using the Sequoia supercomputer managed by LLNL.
The lab will also be provided with an ecosystem developed along with TrueNorth, which will aid them in creating other brain-like computers. What exactly does one need in order to build a neurosynaptic system? The ecosystem kit includes a simulator, a programming language, an integrated programming environment, a library of algorithms and applications, firmware, deep learning tools for neural networks, a curriculum used to teach researchers how to use the system, and cloud enablement. Scientists at LLNL will also be reaching out to the Department of Energy and universities regarding their use of the TrueNorth chip and its ecosystem.
TrueNorth originated in research conducted by the Defense Advanced Research Projects Agency’s (DARPA) Systems of Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE) program, in partnership with Cornell University.
