Scientists Have Created an Artificial Synapse That Can Learn Autonomously

Sergey Tarasov/Shutterstock

Developments and advances in artificial intelligence (AI) have been due in large part to technologies that mimic how the human brain works. In the world of information technology, such AI systems are called neural networks.

These contain algorithms that can be trained, among other things, to imitate how the brain recognises speech and images. However, running an Artificial Neural Network consumes a lot of time and energy.

Now, researchers from the National Centre for Scientific Research (CNRS) in Thales, the University of Bordeaux in Paris-Sud, and Evry have developed an artificial synapse called a memristor directly on a chip.

It paves the way for intelligent systems that required less time and energy to learn, and it can learn autonomously.

In the human brain, synapses work as connections between neurons. The connections are reinforced and learning is improved the more these synapses are stimulated.

The memristor works in a similar fashion. It's made up of a thin ferroelectric layer (which can be spontaneously polarised) that is enclosed between two electrodes.

Using voltage pulses, their resistance can be adjusted, like biological neurons. The synaptic connection will be strong when resistance is low, and vice-versa.

(a) Sketch of pre- and post-neurons connected by a synapse. The synaptic transmission is modulated by the causality (Δt) of neuron spikes. (b) Sketch of the ferroelectric memristor where a ferroelectric tunnel barrier of BiFeO3 (BFO) is sandwiched between a bottom electrode of (Ca,Ce)MnO3 (CCMO) and a top submicron pillar of Pt/Co. YAO stands for YAlO3. (c) Single-pulse hysteresis loop of the ferroelectric memristor displaying clear voltage thresholds ( and ). (d) Measurements of STDP in the ferroelectric memristor. Modulation of the device conductance (ΔG) as a function of the delay (Δt) between pre- and post-synaptic spikes. Seven data sets were collected on the same device showing the reproducibility of the effect. The total length of each pre- and post-synaptic spike is 600 ns.
Source: Nature Communications

The memristor's capacity for learning is based on this adjustable resistance.

AI systems have developed considerably in the past couple of years. Neural networks built with learning algorithms are now capable of performing tasks which synthetic systems previously could not do.

For instance, intelligent systems can now compose music, play games and beat human players, or do your taxes. Some can even identify suicidal behaviour, or differentiate between what is lawful and what isn't.

This is all thanks to AI's capacity to learn, the only limitation of which is the amount of time and effort it takes to consume the data that serve as its springboard.

With the memristor, this learning process can be greatly improved. Work continues on the memristor, particularly on exploring ways to optimise its function.

For starters, the researchers have successfully built a physical model to help predict how it functions.

Their work is published in the journal Nature Communications.

ORIGINAL: ScienceAlert

DOM GALEON, FUTURISM

7 APR 2017

New Memristor Circuit Mimicks Synapses in the Brain

Illustration: University of Massachusetts, Amherst

To a human brain, picking one particular image out of a thousand is an easy task. Billions of neurons and the synapses that connect them can quickly process information in parallel to make a decision. Seeking to make such processing that easy for machines, scientists and engineers have been working with devices called memristors, which have some similar behaviors to neural synapses. Engineers at the University of Massachusetts report this week that they’ve invented a memristor circuit that matches a synapses’ behavior more closely than any before.

First predicted in 1971 and invented in 2008 by HP, memristors are so named because they remember how much voltage you applied across the device and how long you applied it, storing the information as a change in resistance. HP engineers noticed immediately that such a characteristic was similar to the way the synaptic connection between neurons strengthens with use to form a memory.

Memristors are “kind of the ideal candidate in many aspect” says Jianhua Yang, an electrical engineer at the University of Massachusetts Amherst working on improving their behavior. Low-power, simple circuits of memristors could improve the ability of computers to solve power-intensive computer vision and machine learning tasks that human brains handle with little effort.

In research published in Nature Materials on 26 September, Yang and his team mimicked a crucial underlying component of how synaptic connections get stronger or weaker: the flow of calcium.

The movement of calcium into or out of the neuronal membrane, neuroscientists have found, directly affects the connection. Chemical processes move the calcium in and out— triggering a long-term change in the synapses’ strength. 2015 research in ACS NanoLetters and Advanced Functional Materials discovered that types of memristors can simulate some of the calcium behavior, but not all.

In the new research, Yang combined two types of memristors in series to create an artificial synapse. The hybrid device more closely mimics biological synapse behavior—the calcium flow in particular, Yang says.

The new memristor used--called a diffusive memristor because atoms in the resistive material move even without an applied voltage when the device is in the high resistance state—was a dialectric film sandwiched between Pt or Au electrodes. The film contained Ag nanoparticles, which would play the role of calcium in the experiments.

By tracking the movement of the silver nanoparticles inside the diffusive memristor, the researchers noticed a striking similarity to how calcium functions in biological systems.

A voltage pulse to the hybrid device drove silver into the gap between the diffusive memristor’s two electrodes–creating a filament bridge. After the pulse died away, the filament started to break and the silver moved back— resistance increased.

Like the case with calcium, a force made silver go in and a force made silver go out.

To complete the artificial synapse, the researchers connected the diffusive memristor in series to another type of memristor that had been studied before.

When presented with a sequence of voltage pulses with particular timing, the artificial synapse showed the kind of long-term strengthening behavior a real synapse would, according to the researchers. “We think it is sort of a real emulation, rather than simulation because they have the physical similarity,” Yang says.

He says the next step is to better understand the mechanisms. Then, his team plans to combine the artificial synapses into arrays and eventually build bio-inspired circuits.

“We can have a more direct, more natural, and a more complete emulation to the synaptic system using the silver-based system,” he says.

ORIGINAL: IEEE Spectrum

By Andrew Silver

Posted 29 Sep 2016