Learning transformed product distributions

We consider the problem of learning an unknown product distribution over using samples where is a \emph{known} transformation function. Each choice of a transformation function specifies a learning problem in this framework. Information-theoretic arguments show that for every transformation function the corresponding learning problem can be solved to accuracy , using examples, by a generic algorithm whose running time may be exponential in We show that this learning problem can be computationally intractable even for constant and rather simple transformation functions. Moreover, the above sample complexity bound is nearly optimal for the general problem, as we give a simple explicit linear transformation function with integer weights and prove that the corresponding learning problem requires samples. As our main positive result we give a highly efficient algorithm for learning a sum of independent unknown Bernoulli random variables, corresponding to the transformation function . Our algorithm learns to -accuracy in poly time, using a surprising poly number of samples that is independent of We also give an efficient algorithm that uses samples but has running time that is only
View on arXiv