Almost-lossless compression of a low-rank random tensor

TL;DR

This paper investigates the theoretical limits of compressing low-rank random tensors with finite alphabets, focusing on almost-lossless compression and establishing asymptotic bounds.

Contribution

It provides the first asymptotic limit for almost-lossless compression of low-rank random tensors with finite alphabets.

Findings

Established asymptotic compression limits for low-rank tensors

Characterized the conditions for near-perfect tensor reconstruction

Extended understanding of tensor compression in information theory

Abstract

In this work, we establish an asymptotic limit of almost-lossless compression of a random, finite alphabet tensor which admits a low-rank canonical polyadic decomposition.