WHO can see bet­ter?

Forbes - - VERTICALS -

deep-learn­ing sys­tems typ­i­cally iden­tify an im­age by guess­ing the cat­e­gories to which it might be­long—and adding a mea­sure of their con­fi­dence in the guesses. Here are some of what ai sys­tems from clar­i­fai, google, IBM and Mi­crosoft saw in three im­ages we se­lected. (un­like the other three, Mi­crosoft as­signs a con­fi­dence per­cent­age to the “cap­tion” it writes for each im­age while also in­clud­ing a num­ber of tags.)

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.