Memristors Helping to Build a Better Computer Brain
Computers are still quite a ways from mimicking and being just as efficient as the human brain. Our minds are capable of learning and adapting to situations without programming, storing limitless amounts of data, and making computations at fast speeds. It’s also much more efficient than any laptop or desktop (try 20 Watts). The brain is a wonder of computational power, and engineers want to replicate it by creating a better neural network.
Mark Hersam from Northwestern University’s McCormick School of Engineering explained the idea in a press release:
“Computers are very impressive in many ways, but they’re not equal to the mind. Neurons can achieve very complicated computation with very low power consumption compared to a digital computer.”
A team of engineers, including Hersam, have recently taken a step forward in creating a computer that mimics the human brain. Enter memristors, a resistor with the ability to remember (in a manner of speaking).
“Memristors could be used as a memory element in an integrated circuit or computer. Unlike other memories that exist today in modern electronics, memristors are stable and remember their state even if you lose power.”
For the nontechnical, modern devices today use a kind of memory known as RAM, or random access memory. Think back to the specs you were scoping out for your last computer — did it tout having 2GB or 4GB of RAM, or memory? This kind of memory helps speed up multitasking, but when a computer loses power, everything you were doing up to the last auto-save goes down with it. But memristors could provide users with the best of both worlds: a fast and reliable form of memory.
What Hersam and his team have developed is a better kind of memristor, compared to the previous design. They’ve upgraded from the two-terminal design, featuring an input and an output. There’s now another input — a three-terminal device, which allows for more complex systems and allows developers to have more control over the flow of electricity.
“With a memristor that can be tuned with a third electrode, we have the possibility to realize a function you could not previously achieve. A three-terminal memristor has been proposed as a means of realizing brain-like computing. We are now actively exploring this possibility in the laboratory.”
Read more at Science Daily.
Photo Credit: Tomi Knuutila/Flickr
Stefan Weitz, Microsoft director of search, explained how the future of machine and AI advancement will be dependent upon its ability to learn to identify patterns: