Memristors can maybe learn like synapses

Memristors can potentially learn like synapses and be used to build human brain-like computers according to Wei Lu, a University of Michigan scientist.

He thinks that two CMOS circuits connected by a memristor is analogous to two neurons in the brain connected by a synapse. It is thought that synaptic connections strengthen as the neurons either side fire and so brain 'circuits' are established, which constitutes the basis of human learning.

A memristor is resistant to electricity and retains its electrical state when power is switched off - a memory resistor. It can be made more or less resitive by putting a voltage across it. The resistance level is retained when power is switched off.

Lu thinks that cats can process information faster than supercomputers because they process data in parallel instead of in tens of thousand linear streams as in a modern supercomputer - though this seems a strained analogy because the tens of thousands of cores in a supercomputer do act in parallel.

He says a cat brain can recognise a face 83 times faster than a supercomputer because, he posits, synapses are switches that can interconnect thousands of neurons and can change their state to set up and modify brain circuits, such as those involved in facial recognition.

Lu says that a supercomputer, although it can have tens of thousands of processing cores, has little connectivity between them, and so is only good for crunching the data involved in simple tasks with a limited set of variables. This again seems a little forced to us; simulating an atomic explosion is a simple task with limited variables?

But let's suppose that Lu is right and press on. He's connected two CMOS curcuits by a silver and silicon Memristor and powered the two CMOS circuits on and off with varying time gaps between them. The memristor alters its state differently depending on the timing of the powering of the CMOS circuits.

This is said to be the same behaviour as that shown by synapses, called "spike timing plastic dependency", which is thought to be the possible basis for memory and learning in human and other mammalian brains.

The synaptic connection between neurons becomes stronger or weaker, as the time gap between when they are stimulated becomes shorter or longer. In the same way, the shorter the time interval the lower the resistance of the memristor to electricity flowing across it between the two CMOS circuits.

A 20 millisecond time interval between the two CMOS circuits caused a resistance level roughly half that of a 40 millisecond gap. Lu said: "Cells that fire together wire together... The memristor mimics synaptic action.

"We show that we can use voltage timing to gradually increase or decrease the electrical conductance in this memristor-based system. In our brains, similar changes in synapse conductance essentially give rise to long term memory."

He now wants to build prototype devices with tens of thousands of such CMOS circuit/memristor elements, and see if the collection can start responding to stimuli, such as an image of a face, in coherent and dependable ways.

There's more information here and here. Lu's paper or letter is published in NanoLetters and is accessible for a fee. ®