English

Coding sets with asymmetric information

Data Structures and Algorithms 2018-07-30 v2 Information Theory Networking and Internet Architecture math.IT

Abstract

We study the following one-way asymmetric transmission problem, also a variant of model-based compressed sensing: a resource-limited encoder has to report a small set SS from a universe of NN items to a more powerful decoder (server). The distinguishing feature is asymmetric information: the subset SS is comprised of i.i.d. samples from a prior distribution μ\mu, and μ\mu is only known to the decoder. The goal for the encoder is to encode SS obliviously, while achieving the information-theoretic bound of SH(μ)|S| \cdot H(\mu), i.e., the Shannon entropy bound. We first show that any such compression scheme must be {\em randomized}, if it gains non-trivially from the prior μ\mu. This stands in contrast to the symmetric case (when both the encoder and decoder know μ\mu), where the Huffman code provides a near-optimal deterministic solution. On the other hand, a rather simple argument shows that, when S=k|S|=k, a random linear code achieves near-optimal communication rate of about kH(μ)k\cdot H(\mu) bits. Alas, the resulting scheme has prohibitive decoding time: about (Nk)(N/k)k{N\choose k} \approx (N/k)^k. Our main result is a computationally efficient and linear coding scheme, which achieves an O(lglgN)O(\lg\lg N)-competitive communication ratio compared to the optimal benchmark, and runs in poly(N,k)\text{poly}(N,k) time. Our "multi-level" coding scheme uses a combination of hashing and syndrome-decoding of Reed-Solomon codes, and relies on viewing the (unknown) prior μ\mu as a rather small convex combination of uniform ("flat") distributions.

Keywords

Cite

@article{arxiv.1707.04875,
  title  = {Coding sets with asymmetric information},
  author = {Alexandr Andoni and Javad Ghaderi and Daniel Hsu and Dan Rubenstein and Omri Weinstein},
  journal= {arXiv preprint arXiv:1707.04875},
  year   = {2018}
}