Create and explore imaginative 3D worlds with our new product Dreamfield (opens in a new tab)
Papers
Topics
Authors
Recent
Search
2000 character limit reached

Learning Sums of Independent Random Variables with Sparse Collective Support

Published 18 Jul 2018 in cs.DS, cs.LG, math.ST, and stat.TH | (1807.07013v2)

Abstract: We study the learnability of sums of independent integer random variables given a bound on the size of the union of their supports. For A⊂Z+\mathcal{A} \subset \mathbf{Z}_{+}, a sum of independent random variables with collective support A\mathcal{A}} (called an A\mathcal{A}-sum in this paper) is a distribution S=X1+⋯+XN\mathbf{S} = \mathbf{X}_1 + \cdots + \mathbf{X}_N where the Xi\mathbf{X}_i's are mutually independent (but not necessarily identically distributed) integer random variables with ∪isupp(Xi)⊆A.\cup_i \mathsf{supp}(\mathbf{X}_i) \subseteq \mathcal{A}. We give two main algorithmic results for learning such distributions: 1. For the case ∣A∣=3| \mathcal{A} | = 3, we give an algorithm for learning A\mathcal{A}-sums to accuracy ϵ\epsilon that uses poly(1/ϵ)\mathsf{poly}(1/\epsilon) samples and runs in time poly(1/ϵ)\mathsf{poly}(1/\epsilon), independent of NN and of the elements of A\mathcal{A}. 2. For an arbitrary constant k≥4k \geq 4, if A={a1,...,ak}\mathcal{A} = \{ a_1,...,a_k\} with 0≤a1<...<ak0 \leq a_1 < ... < a_k, we give an algorithm that uses poly(1/ϵ)⋅log⁡log⁡ak\mathsf{poly}(1/\epsilon) \cdot \log \log a_k samples (independent of NN) and runs in time poly(1/ϵ,log⁡ak).\mathsf{poly}(1/\epsilon, \log a_k). We prove an essentially matching lower bound: if ∣A∣=4|\mathcal{A}| = 4, then any algorithm must use Ω(log⁡log⁡a4)\Omega(\log \log a_4) samples even for learning to constant accuracy. We also give similar-in-spirit (but quantitatively very different) algorithmic results, and essentially matching lower bounds, for the case in which A\mathcal{A} is not known to the learner.

Citations (3)

Summary

No one has generated a summary of this paper yet.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Continue Learning

We haven't generated follow-up questions for this paper yet.