Janus: Collaborative Vision Transformer Under Dynamic Network Environment

TL;DR

Janus is a novel framework enabling low-latency, collaborative Vision Transformer inference on cloud and edge devices over dynamic networks, balancing accuracy, latency, and communication costs.

Contribution

It introduces a dynamic, collaborative ViT inference method combining token pruning and model splitting to optimize performance under fluctuating network conditions.

Findings

Increases throughput by up to 5.15 times

Reduces latency violation ratios by up to 98.7%

Balances accuracy and latency effectively

Abstract

Vision Transformers (ViTs) have outperformed traditional Convolutional Neural Network architectures and achieved state-of-the-art results in various computer vision tasks. Since ViTs are computationally expensive, the models either have to be pruned to run on resource-limited edge devices only or have to be executed on remote cloud servers after receiving the raw data transmitted over fluctuating networks. The resulting degraded performance or high latency all hinder their widespread applications. In this paper, we present Janus, the first framework for low-latency cloud-device collaborative Vision Transformer inference over dynamic networks. Janus overcomes the intrinsic model limitations of ViTs and realizes collaboratively executing ViT models on both cloud and edge devices, achieving low latency, high accuracy, and low communication overhead. Specifically, Janus judiciously combines token pruning techniques with a carefully designed fine-to-coarse model splitting policy and non-static mixed pruning policy. It attains a balance between accuracy and latency by dynamically selecting the optimal pruning level and split point. Experimental results across various tasks demonstrate that Janus enhances throughput by up to 5.15 times and reduces latency violation ratios by up to 98.7% when compared with baseline approaches under various network environments.