How biological neurons could be the basis for a reservoir computer

Japanese scientists have created an artificial intelligence system that simulates the brain. This is done using a technique called “reservoir calculation”. The researchers used neurons from the rat cerebral cortex to form an artificial “brain”. They found that such a system has a short memory and can classify sequences of data, such as spoken digits. Research results published in Proceedings of the National Academy of Sciences.

The brain is a complex system of billions of interconnected neurons that process information and allow us to perceive the world around us and control our bodies. We still do not fully understand how the structure of the brain and the properties of neurons affect information processing.

Reservoir computing is a technique that mimics how the brain works. It uses a large number of interconnected nodes to transform the input into a more complex form.

The researchers used this method to create an artificial “brain” from rat cortex neurons. Using optogenetics and calcium fluorescence imaging, they were able to record the responses of an artificial neural network. They then processed the data using reservoir computing and found that the artificial brain has a short memory that can be used to classify sequences of data.

It was also found that the artificial brain can classify spoken numbers even when the speakers change. The study proves that biological neurons can serve as a filter to improve reservoir computing.

As a result, the model trained on a particular dataset successfully classified the new dataset as belonging to the same category. This demonstrates the ability of an artificially grown brain to filter information, which increases the efficiency of reservoir computing.