Deep Randomized Distributed Function Computation (DeepRDFC): Neural Distributed Channel Simulation

TL;DR

This paper introduces DeepRDFC, a neural network-based approach for distributed function computation that efficiently simulates target distributions with minimal communication, especially under limited shared randomness.

Contribution

It proposes a deep learning framework using autoencoders for randomized distributed function computation, achieving high performance and communication efficiency.

Findings

Significant reduction in communication load compared to traditional data compression.

Effective distribution simulation with limited common randomness.

Enhanced guarantees in function computation accuracy.

Abstract

The randomized distributed function computation (RDFC) framework, which unifies many cutting-edge distributed computation and learning applications, is considered. An autoencoder (AE) architecture is proposed to minimize the total variation distance between the probability distribution simulated by the AE outputs and an unknown target distribution, using only data samples. We illustrate significantly high RDFC performance with communication load gains from our AEs compared to data compression methods. Our designs establish deep learning-based RDFC methods and aim to facilitate the use of RDFC methods, especially when the amount of common randomness is limited and strong function computation guarantees are required.