A time-space lower bound for a large class of learning problems

We prove a general time-space lower bound that applies for a large class of learning problems and shows that for every problem in that class, any learning algorithm requires either a memory of quadratic size or an exponential number of samples. As a special case, this gives a new proof for the time-space lower bound for parity learning [R16].

Our result is stated in terms of the norm of the matrix that corresponds to the learning problem. Let X, A be two finite sets. Let M:A×X{1,1} be a matrix. The matrix M corresponds to the following learning problem: An unknown element xX was chosen uniformly at random. A learner tries to learn x from a stream of samples, (a1,b1),(a2,b2), where for every i, aiA is chosen uniformly at random and bi=M(ai,x).

Let σ be the largest singular value of M and note that always σ|A|1/2|X|1/2. We show that if σ|A|1/2|X|1/2ϵ, then any learning algorithm for the corresponding learning problem requires either a memory of size quadratic in ϵn or number of samples exponential in ϵn, where n=log2|X|.

Date

Affiliation

Princeton University