Key parameters to improve artificial synaptic devices

Key parameters to improve artificial synaptic devices

The disadvantage of high power consumption of the conventional von Neumann computing method has been overcome by the development of neuromorphic computing systems, which mimic the human brain.

Concept image for article. Image credit: Korea Institute of Science and Technology (KIST).

Implementing a semiconductor device that uses brain information transmission technique requires a high-performance analog artificial synapse device that can express different synaptic connection strengths. This technique takes advantage of the signals sent when a neuron produces a spike signal.

When considering traditional variable resistance memory devices frequently used as artificial synapses, the electric field builds as the filament grows with fluctuating resistance, generating a feedback phenomenon that leads to rapid growth of the filament.

As a result, it is difficult to incorporate a lot of plasticity and maintain an analog (gradual) resistance fluctuation depending on the type of filament.

Dr. YeonJoo Jeong’s team at the Center for Neuromorphic Engineering at the Korea Institute of Science and Technology has overcome long-standing problems with analog synaptic characteristics, plasticity, and information preservation in memristors and devices. neuromorphic semiconductors.

He declared the creation of a synthetic synaptic semiconductor device capable of extremely reliable neuromorphic computation.

The performance of current neuromorphic semiconductor devices has been hampered by modest synaptic plasticity, which the KIST research team addressed by adjusting the redox characteristics of active electrode ions.

Additionally, several transition metals have been doped and used in the synaptic device to alter the reduction probability of active electrode ions. The high probability of ion reduction has proven to be a crucial factor in the creation of high-performance artificial synaptic devices.

The study team therefore added a titanium transition metal with a high ion reduction probability to an already existing artificial synaptic device.

This preserves the analog properties of the synapse and the plasticity of the device at the biological brain synapse, which is about five times the difference between high and low resistance.

The team also created a high-performance neuromorphic semiconductor that is about 50 times more efficient.

Compared with the current artificial synaptic device, the information retention was improved up to 63 times due to the high alloying reaction involving the titanium-doped transition metal. More precise simulations of certain brain processes, such as depression and long-term potentiation, could also be performed.

Using the artificial synaptic device they had built, the team attempted to implement an artificial neural network learning model for image recognition. As a result, compared with the current artificial synaptic device, the error rate has been reduced by more than 60%.

Additionally, handwriting image pattern recognition (MNIST) accuracy increased by more than 69%. By improving the artificial synaptic device, the research team demonstrated the viability of a high-performance neuromorphic computing system.

This study significantly improved synaptic range of motion and information preservation, which were the biggest technical hurdles of existing synaptic mimics. In the developed artificial synapse device, the analog operation area of ​​the device to express the different connection strengths of the synapse has been maximized, so that the performance of artificial intelligence based on brain simulation will be improved.

Dr. YeonJoo Jeong, Senior Researcher, Neuromorphic Engineering Center, Korea Institute of Science and Technology

Dr Jeong added: “In the follow-up research, we will fabricate a neuromorphic semiconductor chip based on the developed artificial synapse device to realize a high-performance artificial intelligence system, thus improving the competitiveness in the home system and semiconductor field of artificial intelligence.

Journal reference

Kang, J. et al. (2022) Cluster-like analog memristor by redox dynamics engineering for high-performance neuromorphic computing. Communication Nature. doi: 10.1038/s41467-022-31804-4.


#Key #parameters #improve #artificial #synaptic #devices

Leave a Comment

Your email address will not be published.