Sign In

Communications of the ACM

ACM News

Brain Connectivity Can Build Better AI


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
Brain Connectivity Can Build Better AI

Using artificial networks can help to understand how brain structure supports brain function, while using empirical data to make neural networks can reveal design principles for building better AI.

A new study shows that artificial intelligence (AI) networks based on human brain connectivity can efficiently perform cognitive tasks.

Examining MRI data from a large Open Science repository, a team of researchers from The Neuro (Montreal Neurological Institute-Hospital) and the Quebec Artificial Intelligence Institute reconstructed a brain connectivity pattern and applied it to an artificial neural network (ANN). They trained the ANN to perform a cognitive memory task and observed how it worked to complete the assignment, hoping to both learn how the wiring of the brain supports specific cognitive skills and to derive novel design principles for artificial networks.

Researchers found that ANNs with human brain connectivity, known as neuromorphic neural networks, performed cognitive memory tasks more flexibly and efficiently than other benchmark architectures.

From McGill University
View Full Article


 

No entries found