Editing Eye tracking (section)

=== Game theory applications ===
In a 2019 study, a Convolutional Neural Network (CNN) was constructed with the ability to identify individual chess pieces the same way other CNNs can identify facial features.<ref name=":1">{{Cite book|last1=Louedec|first1=Justin Le|last2=Guntz|first2=Thomas|last3=Crowley|first3=James L.|last4=Vaufreydaz|first4=Dominique|title=Proceedings of the 11th ACM Symposium on Eye Tracking Research & Applications |chapter=Deep learning investigation for chess player attention prediction using eye-tracking and game data |date=2019|pages=1–9|location=New York, New York, USA|publisher=ACM Press|doi=10.1145/3314111.3319827|isbn=978-1-4503-6709-7|arxiv=1904.08155|bibcode=2019arXiv190408155L|s2cid=118688325}}</ref> It was then fed eye-tracking input data from 30 chess players of various skill levels. With this data, the CNN used gaze estimation to determine parts of the chess board to which a player was paying close attention. It then generated a saliency map to illustrate those parts of the board. Ultimately, the CNN would combine its knowledge of the board and pieces with its saliency map to predict the players' next move. Regardless of the [[Training data set|training dataset]] the neural network system was trained upon, it predicted the next move more accurately than if it had selected any possible move at random, and the saliency maps drawn for any given player and situation were more than 54% similar.<ref name=":1" />