The chips are able to mimic important aspects of biological brains by being energy efficient, resilient and able to learn. These chips promise to have a major impact on the future of artificial intelligence.

Picture shows a computer chip

The new BrainScaleS chip has an improved ability to reproduce the continual process of learning, compared to the previous generation of the chip. The new generation SpiNNaker chip features an improved power management for very efficient energy usage that makes it suitable for real-time simulation of multi-scale brain models.

“We have understood that learning is the key for all applications of neuromorphic systems. The new architectures are optimized for fast and efficient learning. In the Human Brain Project (HBP) this important development has been carried out in close collaboration with theoretical neuroscientists. This idea of a co-design process was the key for today’s achievement and it makes us all very proud", says Prof. Dr. Karlheinz Meier, who leads the area of neuromorphic computing of the Human Brain Project. Neuromorphic computing implements aspects of biological neural networks as analogue or digital copies on electronic circuits. The goals of neuromorphic computing are offering a tool for neuroscience to understand the dynamic processes of learning and development in the brain and applying brain inspiration to generic cognitive computing.

The two new chips were presented end of February at the "Neuro inspired computational elements (NICE)" workshop in the United States, together with another neuromorphic chip from Intel Corporation.

More information