JMSNAS: Joint Model Split and Neural Architecture Search for Learning over Mobile Edge Networks

TL;DR

This paper introduces JMSNAS, a framework that automatically generates and deploys neural networks over mobile edge networks by jointly optimizing model splitting and architecture search considering resource constraints.

Contribution

It proposes a novel joint model split and neural architecture search method tailored for mobile edge networks, addressing resource constraints and latency-accuracy trade-offs.

Findings

Outperforms state-of-the-art split machine learning methods.

Effectively balances accuracy and latency in edge deployments.

Automates model generation and splitting for resource-constrained environments.

Abstract

The main challenge to deploy deep neural network (DNN) over a mobile edge network is how to split the DNN model so as to match the network architecture as well as all the nodes' computation and communication capacity. This essentially involves two highly coupled procedures: model generating and model splitting. In this paper, a joint model split and neural architecture search (JMSNAS) framework is proposed to automatically generate and deploy a DNN model over a mobile edge network. Considering both the computing and communication resource constraints, a computational graph search problem is formulated to find the multi-split points of the DNN model, and then the model is trained to meet some accuracy requirements. Moreover, the trade-off between model accuracy and completion latency is achieved through the proper design of the objective function. The experiment results confirm the superiority of the proposed framework over the state-of-the-art split machine learning design methods.