Scientists from the University of Carnegie Mellon in Pittsburgh (USA) have developed a deep learning algorithm that is able to analyze brain scintigraphy and guess what a person thinks. Interpretation of signals sent by the brain allows AI to "decode" even complex sentences.

Researchers "fed" the algorithms of machine learning data collected with a device for functional magnetic resonance imaging (fMRI). Having studied them, AI in a certain sequence built a chain, which uses the brain to create those or other thoughts, hitech.vesti.ru reports.

In total, about 240 brain reactions to individual people, places or physical actions were identified. Comparing the data fMRI with the reactions already known to them, the algorithms tried to guess what the person is thinking about right now. To the surprise of scientists, the accuracy of predictions was very high - up to 87%. Moreover, AI were able to roll back the operation in reverse order, reproducing on the fMRT-image the offer received from the person.

© All rights reserved. Citing to www.ict.az is necessary upon using news