Wednesday, October 28

ai

ai, Apps & Games, News, tennis

Researchers remake the 2019 Wimbledon final in a realistic AI simulation

In the final of Wimbledon 2019, Roger Federer came one point to take the cup, only to fail twice, allowing Novak Djokovic to get ahead. Now, however, the story has been altered through a new AI-generated simulation that allows Federer to win after all. Researchers from California's Stanford University have trained the AI ​​using a database of images of tennis matches, which enables a statistical model to predict the actions of individual players in certain situations.Vid2Player: Controllable video sprites that behave and look like professional tennis playerspdf: https://t.co/XUHyI6Pp4iABS: https://t.co/5X0Lb2MgAPproject page: https://t.co/wYBTTo9JBi pic.twitter.com/gOh3nEyVKv - AK (@ ak92501) August 12, 2020For example, the model has learned the differences between how Federer and D...
ai, Apps & Games, bias, Montreal, News

Montreal ethics group proposes ways to counter AI model bias

Nonprofit analysis group Montreal AI Ethics Institute has outlined some ways in which firms can sort out inherent bias in AI fashions. In its first-ever State of AI Ethics report, the group notes that rating and suggestion algorithms usually take away topics from their cultural and contextual social meanings. In line with the researchers, builders are sometimes “making an attempt to cut back gender and race into discrete classes which are one-dimensional, third-party, and algorithmically ascribed.” As an alternative of this, the researchers suggest a framework that appears as subjective classes primarily based on a pool of “numerous” people, or “determinantal level course of” (DPP). Primarily, which means that the mannequin will co...