I just came here because of an Linkedin article that explained how much more ad revenue this content makes opposed to the identical one on Youtube. Wanted to see if this video actually exists...
say your model predicts a result and the correct result is something else, the loss function would tell your model how off it was by calculating the difference.
So say your model is supposed to give you something like the sin(x) if you give it x.
You gave it x and it predicted y. The actual result, which you already know (and the model doesn't) is z, then x-z would be the loss, and the loss function would be L = y - z, where y is the prediction and z the actual correct result.
所有评论 (32)
登录或注册马上射条评论!