SL-ACC: A Communication-Efficient Split Learning Framework with Adaptive Channel-wise Compression

2025-08-18

AI Summary

To address the communication bottleneck in split learning (SL) caused by transmitting massive smashed data (activations and gradients) from edge devices, this paper proposes SL-ACC, an efficient grouped compression framework. Methodologically, SL-ACC first quantifies channel importance via Shannon entropy for fine-grained contribution assessment; it then introduces adaptive channel grouping coupled with intra-group differential compression to minimize transmission overhead while preserving model accuracy. Extensive experiments across multiple benchmark datasets demonstrate that SL-ACC significantly accelerates convergence—reducing training time to reach target accuracy compared to state-of-the-art methods—while cutting communication volume by up to 68%. Crucially, accuracy degradation remains bounded at less than 0.5%, confirming robustness. SL-ACC thus delivers a cost-effective, communication-efficient solution for distributed collaborative training under resource-constrained edge environments.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Search and Optimization: Learning to Search
• Computer Vision: Learning & Optimization for CV

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

The increasing complexity of neural networks poses a significant barrier to the deployment of distributed machine learning (ML) on resource-constrained devices, such as federated learning (FL). Split learning (SL) offers a promising solution by offloading the primary computing load from edge devices to a server via model partitioning. However, as the number of participating devices increases, the transmission of excessive smashed data (i.e., activations and gradients) becomes a major bottleneck for SL, slowing down the model training. To tackle this challenge, we propose a communication-efficient SL framework, named SL-ACC, which comprises two key components: adaptive channel importance identification (ACII) and channel grouping compression (CGC). ACII first identifies the contribution of each channel in the smashed data to model training using Shannon entropy. Following this, CGC groups the channels based on their entropy and performs group-wise adaptive compression to shrink the transmission volume without compromising training accuracy. Extensive experiments across various datasets validate that our proposed SL-ACC framework takes considerably less time to achieve a target accuracy than state-of-the-art benchmarks.

Problem

• Reduces communication bottleneck in split learning from excessive data transmission
• Compresses activation and gradient data adaptively without accuracy loss
• Accelerates distributed neural network training on resource-constrained devices

Innovation

• Adaptive channel importance identification using entropy
• Channel grouping compression for smashed data
• Reduces transmission volume without accuracy loss