Summary: Researchers have developed a more powerful and energy-efficient memristor, based on the structure of the human brain, that combines data storage and processing. The new technology, made from nanocrystals of halogenated perovskite, is not yet ready for use as it is difficult to integrate with existing computer chips, but it has the potential for parallel processing of large amounts of data.
Source: Politecnico di Milano
Inspired by the brain’s energy efficiency, copying its structure to create more powerful computers, a team of researchers from Politecnico di Milano, Empa and ETH Zurich has developed a memristor that is more powerful and easier to produce than its predecessors: the results have been published in Science Advances.
The researchers are developing computer architectures inspired by the functioning of the human brain through new components that, like brain cells, combine data storage and processing. The new memristors are based on nanocrystals of halogenated perovskite, a semiconductor material known for the production of solar cells.
Although most people cannot do mathematical calculations with computer precision, humans can effortlessly process complex sensory information and learn from their experiences – a thing that no computer can (yet) do. And in doing so, the human brain consumes just half the energy of a laptop thanks to its structure in synapses, capable of both storing and processing information.
In computers, however, the memory is separate from the processor and data must be continuously transported between these two units. The transport speed is limited and this makes the whole computer slower when the amount of data is very large.
‘Our goal is not to replace the classic computer architecture.’ – explains Daniele Ielmini, professor at Politecnico di Milano – ‘Rather, we want to develop alternative architectures that can perform certain tasks faster and more energy-efficiently. This includes, for example, the parallel processing of large amounts of data; today this happens everywhere, from agriculture to space exploration.’
Based on the measurements, the researchers simulated a complex computational task that corresponds to a learning process in the visual cortex of the brain. The task was to determine the orientation of a light bar based on signals from the retina.
‘Halide perovskites conduct both ions and electrons.’ – clarifies Rohit John, postdoc at ETH Zurich and Empa – ‘This dual conductivity allows for more complex calculations that are more similar to brain processes.’
The technology is not ready for use yet and simply manufacturing the new memristors makes integrating them with existing computer chips difficult: perovskites cannot handle the 400-500 °C temperatures needed for silicon processing – at least not yet.
There are also other materials with similar properties that could be considered for the production of high performance memristors. ‘We can test the results of our memristor system with different materials,’ says Alexander Milozzi, Ph.D candidate at Politecnico di Milano – ‘probably some of them are more suitable for integration with silicon.’
Summary generated by Chat GPT AI technology
About this neurotech research news
Author: Emanuele Sanzone
Source: Politecnico di Milano
Contact: Emanuele Sanzone – Politecnico di Milano
Image: The image is in the public domain
Original Research: Open access.
“Ionic-electronic halide perovskite memdiodes enabling neuromorphic computing with a second-order complexity” by Rohit John et al. Science Advances
Ionic-electronic halide perovskite memdiodes enabling neuromorphic computing with a second-order complexity
With increasing computing demands, serial processing in von Neumann architectures built with zeroth-order complexity digital circuits is saturating in computational capacity and power, entailing research into alternative paradigms.
Brain-inspired systems built with memristors are attractive owing to their large parallelism, low energy consumption, and high error tolerance.
However, most demonstrations have thus far only mimicked primitive lower-order biological complexities using devices with first-order dynamics.
Memristors with higher-order complexities are predicted to solve problems that would otherwise require increasingly elaborate circuits, but no generic design rules exist.
Here, we present second-order dynamics in halide perovskite memristive diodes (memdiodes) that enable Bienenstock-Cooper-Munro learning rules capturing both timing- and rate-based plasticity.
A triplet spike timing–dependent plasticity scheme exploiting ion migration, back diffusion, and modulable Schottky barriers establishes general design rules for realizing higher-order memristors.
This higher order enables complex binocular orientation selectivity in neural networks exploiting the intrinsic physics of the devices, without the need for complicated circuitry.