by Summary: Researchers have developed a more powerful and energy-efficient memristor based on the structure of the human brain, combining data storage and processing. The new technology, made from halogenated perovskite nanocrystals, is not yet ready for use because it is difficult to integrate into existing computer chips, but it has the potential for processing large amounts of data in parallel.
Source: Politecnico di Milano
Inspired by the energy efficiency of the brain, which copies 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 manufacture than its predecessors: the results have been published In scientific advances.
Researchers are developing computer architectures inspired by how the human brain works, through new components that combine data storage and processing like brain cells. The new memristors are based on nanocrystals of halogenated perovskite, a semiconductor material known for making solar cells.
Although most humans cannot perform mathematical calculations with computer precision, humans can effortlessly process complex sensory information and learn from their experiences — something no computer can (yet). The human brain uses only half as much energy as a laptop thanks to its structure of synapses, which can both store and process information.
However, in computers, memory is separate from the processor, and data must be continuously transported between these two units. The transport speed is limited, which makes the whole computer slow down when the amount of data is very large.
“Our goal is not to replace the classic computer architecture.” – explains Daniele Ielmini, Professor at the 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 is happening everywhere, from agriculture to space exploration.’
Using the measurements, the researchers simulated a complex arithmetic task that corresponds to a learning process in the brain’s visual cortex. The task was to determine the orientation of a light bar based on signals from the retina.
“Halide perovskites conduct both ions and electrons.” – explains Rohit John, postdoc at ETH Zurich and Empa – “This double conductivity enables more complex calculations that are more similar to brain processes.”
The technology isn’t ready yet, and the new memristors’ ease of manufacture makes them difficult to integrate into existing computer chips: perovskites can’t handle the 400 to 500°C temperatures required for silicon processing — at least not yet.
There are also other materials with similar properties that could be considered for making high performance memristors. “We can test the results of our memristor system with different materials,” says Alexander Milozzi, PhD student at the Politecnico di Milano – “probably some of them are better suited for integration with silicon.”
Summary generated by Chat GPT AI technology
About this news from neurotech research
Author: Emanuele Sanzone
Source: Politecnico di Milano
Contact: Emanuele Sanzone – Politecnico di Milano
Picture: The image is in the public domain
Original research: Open access.
“Ionic-Electronic Halide Perovskite Memdiodes Enable Second-Order Complexity Neuromorphic Computation” by Rohit John et al. scientific advances
Abstract
Halide perovskite ionic-electronic memdiodes enabling second-order complexity neuromorphic computing
As computational demands increase, serial processing in von Neumann architectures built with zero-order complexity digital circuits saturates in terms of computational capacity and performance, prompting exploration of alternative paradigms.
Brain-inspired systems built with memristors are attractive because of their high parallelism, low power consumption, and high fault tolerance.
However, most demonstrations to date have only mimicked primitive, low-order biological complexities using first-order dynamics devices.
Memristors of higher complexity aim to solve problems that would otherwise require increasingly sophisticated circuits, but there are no generic design rules.
Here we present second-order dynamics in memristive halide perovskite diodes (memdiodes) that enable beehive-Cooper-Munro learning rules that capture both time- and rate-based plasticity.
A triplet spike timing-dependent plasticity scheme using ion migration, back-diffusion, and tunable Schottky barriers establishes general design rules for the realization of higher-order memristors.
This higher order enables complex binocular orientational selectivity in neural networks that exploit the intrinsic physics of the devices without the need for complicated circuitry.
