Home / Science / A new brain-based computing device simulates human learning.

A new brain-based computing device simulates human learning.

A new brain-based computing device simulates human learning.

Researchers show that their device can simulate coherence learning by connecting a single synaptic transistor to a neural circuit.credit: Northwestern University.

Researchers have developed a brain-like computer device that can be learned by making connections.

Like how famous physiologist Ivan Pavlov has reconditioned dogs to link their bells to food, researchers at Northwestern University and the University of Hong Kong have been successful in modulating their cycles to link light to pressure.

The research will be published in the journal April 30. Nature communication.

The secret of the device is in Electrochemical organic “synaptic transistors”

; process and store information at the same time as the human brain. The researchers showed that transistors could mimic the short- and long-term flexibility of the synapse in the human brain, building on memories to learn over time.

With brain-like capabilities, new transistors and circuits may overcome the limitations of traditional computing, including power-efficient hardware and limited ability to multitask. Brain-like devices also have a higher fault tolerance and continue to operate smoothly even if some components fail.

“Although modern computers are unique But the human brain can perform better in complex and unstructured tasks such as pattern recognition, motor control and multi-memory integration, ”said Jonathan Rivnay of Northwestern, the senior author of the study. “This is thanks to the plasticity of synapse, which is a fundamental component of the brain’s computational power. We, we demonstrate Organic plastic transistors that mimic the core function of a biological synapse.

Rivnay is an assistant professor of biomedical engineering at Northwestern’s McCormick School of Engineering.He led the study together with Paddy Chan, an associate professor of mechanical engineering at Hong Kong University.Xudong Ji, a postdoctoral researcher in Rivnay’s group, is the author of the paper. first

Problems with legacy processing

A typical digital computer system has a separate processor and storage unit, making data-intensive tasks use a lot of energy. In recent years, inspired by computing and data storage in the human brain, it has sought to develop computers that function more similarly to the human brain, with arrays of devices acting like networks. Of neurons

“The way computer systems work today is that memory and logic are separate,” said G. If we can combine these two separate functions, we can save space and save on energy bills. “

A new brain-based computing device simulates human learning.

Researchers show that their device can simulate coherence learning by connecting a single synaptic transistor to a neural circuit.credit: Northwestern University.

Memory resistor or “Memister” is the best developed technology that can work together in processing and memory. But Memphis suffered from high-cost energy conversion and less biocompatibility. These disadvantages led the researchers to synaptic transistors, particularly electrochemical synaptic transistors, which operate at low voltages, continuously tunable memory, and compatibility. High for biological applications There are still challenges.

“Even high-performance organic, electrochemical, synaptic transistors require write operations to separate them from the read operations,” Rivnay said. From the writing process, which can complicate integration into a circuit or system. “

How do Synaptic transistors work?

To overcome these challenges, the Northwestern team and the University of Hong Kong optimize the conductive plastic materials within the organic electrochemical transistors that can trap ions. In the brain, synapse is a structure in which neurons can send signals to other neurons using small molecules called neurotransmitters. In transistors, the synaptic ions work like a neurotransmitter, transmitting signals between the terminals to form an artificial synapse. By preserving the stored information from the trapped ions, the transistor remembers previous activity, which is the long-term development of plasticity.

The researchers demonstrated the device’s synaptic behavior by connecting a single synaptic transistor to a neuron circuit to simulate cross-link learning. They integrate pressure and light sensors into the circuit and train the circuit to link two unrelated physical inputs. (Pressure and light) together

Perhaps the most famous example of associative learning is Pavlov’s dog, who naturally drool when encountering food. After reconditioning the dog to connect the buzzer to food, the dog began to drool when the bell sounded for the neuromorphic circuit, the researchers activated a voltage using their fingers. To recondition the circuit to link light with pressure, the researchers first applied pulsed light from an LED lamp and then applied pressure immediately. In this situation, the pressure is food and the light is the bell. The respective sensors of the device detected both inputs.

After one training cycle, a connection was made between the light and the initial pressure. After five training cycles, the cycle is significantly associated with pressure. Light alone can stimulate a signal or “Unconditional response”

Future use

Since the synaptic circuits are made from soft polymers like plastic, they can be fabricated on flexible sheets and combined with soft, wearable electronic devices. And implantation devices that are directly connected to living tissue and even the brain

“Although our application is a proof of concept But our proposed circuit can be further expanded to integrate sensory inputs and combine them with other electronic devices to enable on-site computing and low power consumption, ”Rivnay said. Therefore, the device can be directly connected to the living tissue, which is critical for the next generation of electronics. ”

Invention synapse inspired by mechanism-photonics.

More information:
“Coherent learning mimics using non-volatile organic electrochemical synaptic transistors” Nature communication (2021). DOI: 10.1038 / s41467-021-22680-5

Provided by Northwestern University

Reference: Brain-computer-virtual device to simulate human learning (2021, April 30) .Retrieved April 30, 2021 from https://techxplore.com/news/2021-04-brain-like-device-simulate-human.html.

This document is copyrighted. In addition to fair agreements for personal study or research purposes, no part may be reproduced without written permission. The content is for information purposes only.

Source link