FrankenSplit: Efficient Neural Feature Compression with Shallow Variational Bottleneck Injection for Mobile Edge Computing

TL;DR

This paper introduces FrankenSplit, a neural feature compression method optimized for mobile edge computing that significantly reduces data transmission size and speeds up processing without sacrificing accuracy.

Contribution

It presents a novel variational compression framework tailored for resource-constrained edge devices, outperforming existing codecs in efficiency and speed.

Findings

Achieves 60% lower bitrate than state-of-the-art SC methods.

Up to 16x faster than existing codec standards for offloading.

Maintains accuracy while reducing data transmission size.

Abstract

The rise of mobile AI accelerators allows latency-sensitive applications to execute lightweight Deep Neural Networks (DNNs) on the client side. However, critical applications require powerful models that edge devices cannot host and must therefore offload requests, where the high-dimensional data will compete for limited bandwidth. This work proposes shifting away from focusing on executing shallow layers of partitioned DNNs. Instead, it advocates concentrating the local resources on variational compression optimized for machine interpretability. We introduce a novel framework for resource-conscious compression models and extensively evaluate our method in an environment reflecting the asymmetric resource distribution between edge devices and servers. Our method achieves 60% lower bitrate than a state-of-the-art SC method without decreasing accuracy and is up to 16x faster than offloading with existing codec standards.