For noisy targets, are we still trying to learn the deterministic f(x)?

[samuel]

i believe linear regression algorithm can be used for noisy targets, yet it appears to be estimating f(x), I guess the mean of y given X, so how is that helping us to learn P(y|X)?

I believe the professor said that for noisy targets we aim to learn P(y|X) as our target function. Did i misunderstand?

[magdon]

You are still trying to learn , even when there is noise. The case where you are trying to learn is when you are doing logistic regression.