MIT Technology Review reports:
A new form of computer memory might help machines match the capabilities of the human brain when it comes to tasks such as interpreting images or video footage.
Researchers at IBM used what’s known as phase-change memory to build a device that processes data in a way inspired by the workings of a biological brain. Using a prototype phase-change memory chip, the researchers configured the system to act like a network of 913 neurons with 165,000 connections, or synapses, between them. The strength of those connections change as the chip processes incoming data, altering how the virtual neurons influence one another. By exploiting that property, the researchers got the system to learn to recognize handwritten numbers.
Related Stories
A new brain-inspired architecture could improve how computers handle data and advance AI
IBM researchers are developing a new computer architecture, better equipped to handle increased data loads from artificial intelligence. Their designs draw on concepts from the human brain and significantly outperform conventional computers in comparative studies. They report on their recent findings in the Journal of Applied Physics, from AIP Publishing.
[...] The IBM team drew on three different levels of inspiration from the brain. The first level exploits a memory device's state dynamics to perform computational tasks in the memory itself, similar to how the brain's memory and processing are co-located. The second level draws on the brain's synaptic network structures as inspiration for arrays of phase change memory (PCM) devices to accelerate training for deep neural networks. Lastly, the dynamic and stochastic nature of neurons and synapses inspired the team to create a powerful computational substrate for spiking neural networks.
[...] Last year, they ran an unsupervised machine learning algorithm on a conventional computer and a prototype computational memory platform based on phase change memory devices. "We could achieve 200 times faster performance in the phase change memory computing systems as opposed to conventional computing systems." Sebastian said. "We always knew they would be efficient, but we didn't expect them to outperform by this much." The team continues to build prototype chips and systems based on brain-inspired concepts.
Biosensor response from target molecules with inhomogeneous charge localization (DOI: 10.1063/1.5036538) (DX)
Previously: IBM Chip Processes Data Similar to the Way Your Brain Does
IBM Builds New Form of Memory that Could Advance Brain-Inspired Computers
Simulating Neuromorphic Supercomputing Designs
The Second Coming of Neuromorphic Computing
Novel Synaptic Architecture for Brain Inspired Computing
(Score: 0) by Anonymous Coward on Wednesday December 31 2014, @06:02PM
Is this a "property" of phase-change memory or just poor explanation? Could this be done with RRAM or memristors?
(Score: 2) by The Archon V2.0 on Wednesday December 31 2014, @06:58PM
As far as I know PCM can hold more than a binary state in each "bit", but I don't see how that can't be emulated with software and more conventional RAM.
I think something's missing from this story. Any AI or bleeding-edge fanciers here that can enlighten us?
(Score: 0) by Anonymous Coward on Wednesday December 31 2014, @10:53PM
The article specifically said its 1 bit per cell binary storage. They are (implicitly!) comparing it to flash, not RAM, and like it for its ease of writing (compared to flash) and density. From the article:
"Phase-change memory is thought to be particularly well suited to neuromorphic computer systems because it stores data so densely, making it possible to create brain-inspired systems with many more synapses, says Burr. Phase-change memory is also simpler to reprogram. That makes it practical for building a neuromorphic system that is able to “learn” by adjusting its behavior as it is fed new data."
(Score: 2) by frojack on Wednesday December 31 2014, @10:57PM
Even trinary/ternary memory can hold more than on or off, and early workers in this field [wikipedia.org] have made statements suggesting that had ternary been the basis of computing it would be more line the human brain.
Yet I don't see it either.
Humans often like or dislike or are neutral towards things, but the hardly seems a basis to claim an advantage.
No, you are mistaken. I've always had this sig.
(Score: 2) by The Archon V2.0 on Wednesday December 31 2014, @06:02PM
Because I can't read my own handwriting and I'd really like to not have to guess this phone number.
(Score: 2) by francois.barbier on Wednesday December 31 2014, @07:35PM
Written by yourself?
On your hand?
Yesterday night while you where drunk?
There's probably an app for that...
(Score: 2) by darkfeline on Wednesday December 31 2014, @11:52PM
TFS makes it sound like a hardware implementation of a neural network. Is that really noteworthy?
Join the SDF Public Access UNIX System today!
(Score: 1) by klondike0 on Thursday January 01 2015, @05:45PM
I appreciate that somebody has that headache and manages to make it work out. As far as the importance goes, I think the implications of memory "variability" in modern computing deserves more debate -- what are we to do when our digital servants make errors like and unlike humans?