The team develops AI to decode brain signals and predict behavior

Cells in the mouse brain. Credit: Flickr’s ZEISS microscope (CC BY-NC-ND 2.0)

Designed by an international team involving UCL, artificial neural networks (AI) transform raw data from brain activity, paving the way for new discoveries and technologies and closer integration of the brain.

New methods may accelerate the discovery of how brain activity is related to behavior.

Study published today eLifeCo-led by the Kavli Institute for Systems Neuroscience in Trondheim and the Max Planck Institute for Human Cognitive and Brain Sciences Leipzig and funded by the Welcome and European Research Council, Convolutional neural network, Certain types of deep learning algorithmCan decipher different behaviors and stimuli from different brain regions of different species, including humans.

Principal Researcher Markus Frey (Kavli Institute for Systems Neuroscience) said: “Neurologists have been able to record ever-larger datasets from the brain, but they understand the information contained in that data. Nervous code— Still a difficult problem. In most cases, you don’t know which message is being sent.

“We wanted to avoid the need to manually decode different types of raw neural data and develop an automated method to analyze them.”

They tested a network called DeepInsight with neural signals from rats exploring the open arena and found that they could accurately predict animal position, head orientation, and running speed. Even without manual processing, the results were more accurate than those obtained by traditional analysis.

Senior author Professor Caswell Barry (UCL Cell & Developmental Biology) said: Teach us to read the code and in doing so some of those other elements.

“We can decode neural data more accurately than before, but the real advance is that the network is not constrained by existing knowledge.”

The team discovered that their model could identify new aspects of the neural code. This is demonstrated by detecting previously unrecognized head orientation representations encoded by interneurons in the hippocampal region that first showed functional defects in people. With Alzheimer’s disease.

In addition, they show that the same network can predict behavior from different types of records throughout the brain region and can also be used to infer hand movements of human participants. This was determined by testing the network with an existing dataset. Brain activity Recorded by a person.

Co-author Professor Christian Doeller (Kavli Institute for Systems Neuroscience and Max Planck Institute for Human Cognitive and Brain Sciences) said: problem solving. “

“Our framework allows researchers to perform rapid, automated analysis of raw neural data, using traditional methods and spending only on the most promising hypotheses,” said Markus Frey. You can save time. ”

Hide malware in AI neural networks

For more information:
Markus Frey et al, Interpretation of Broadband Neural Activity Using Convolutional Neural Networks, eLife (2021). DOI: 10.7554 / eLife.66551

Journal information:

Quote: The team develops an AI that decodes brain signals and predicts behavior (August 17, 2021).

This document is subject to copyright. No part may be reproduced without written permission, except for fair transactions for personal investigation or research purposes. The content is provided for informational purposes only.

The team develops AI to decode brain signals and predict behavior

Source link The team develops AI to decode brain signals and predict behavior

Show More

Related Articles

Back to top button