Artificial neurons mimic complex brain functions for next-generation AI computing
Researchers have developed artificial neurons that are atomically thin and can process both electric and light signals. The material allows for separate feedforward and backward paths to exist simultaneously within a network of neural neurons, increasing the ability to solve complicated problems.
Scientists have spent decades researching how to replicate the computational versatility of biological neurons in order to create faster and energy-efficient machine-learning systems. Memristors are a promising technology that can store a value and use it for in-memory computing.
The difficulty of integrating feedforward and feedback neuronal signal has proven to be a major challenge in replicating complex biological processes using memristors. These mechanisms are the basis of our ability to learn complex skills using both rewards and mistakes.