Re:Google chief: I'd disclose smart speakers before guests e...
by chenning on Thursday October 17, 2019 @02:20PM
(#21284935)
Don't you need data aggregation to build the model though? I think that's part of the problem, but I could be wrong.
[ Parent
| on HN ]
Re:Google chief: I'd disclose smart speakers before guests e...
by godelski on Thursday October 17, 2019 @04:05PM
(#21285869)
No. You could do this with any model that has feedback built in. I don't think you'd even need something as complicated as Q-learning but any type of reinforcement "learning" method would work. You're basically teaching a PID that has a clock.