Focus-Science and Technology - - DISCOVERIES -


Re­searchers at MIT have cre­ated a neu­ral net­work model that can look for signs of de­pres­sion in recorded hu­man speech and text. They formed the model by feed­ing se­quences of text and speech into the net­work and al­low­ing it to ‘learn’ the tell-tale signs that a per­son was de­pressed. In sub­se­quent tests the model suc­cess­fully picked out de­pressed peo­ple with 77 per cent ac­cu­racy.

Newspapers in English

Newspapers from UK

© PressReader. All rights reserved.