mlsanityData to 7 Oct 2026

Contribution-Aware Bandwidth Allocation for Multimodal Split Learning

arXiv 2609.01406 ↗First posted 1 Sep 2026cs.LG cs.DC cs.NIcited by 0

Not on the map. Why?

Abstract (arXiv)

Multimodal models are increasingly the default option for perception at the network edge, yet they are trained almost entirely in the datacenter, because a client holding several sensor streams cannot host an encoder per modality. Split Learning makes such training feasible by keeping only the first layers on the device, at the cost of an uplink that must carry smashed activations for every modality at every step. Existing compression schemes give each modality the same keep-ratio, so the shared budget is divided in proportion to smashed-activation dimension, a quantity unrelated to how much each modality contributes to the fused prediction. We make that division an explicit decision and call it inter-modality allocation: under a fixed uplink budget, every policy transmits the same expected payload and differs only in how that payload is split across modalities. Our allocator, ModalShare, sets each modality's keep-ratio from a Shapley contribution score that the server computes over coalitions of activations it has already received. Measuring this score adds no uplink traffic and no client-side computation, and needs no prior knowledge of which stream is which. ModalShare improves accuracy over equal keep-ratios by 15.4 and 12.4 percentage points on CREMA-D and MVSA at matched payload in 5x compression, with strong performance across three compressors, three datasets, and four budgets. We show that existing compressors underperform in multimodal settings, with ModalShare recovering what gains are left behind.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 23 of its 33 references that are in the map. First posted on arXiv 1 Sep 2026.

Builds on

23 of 33 references.

References · 23

  1. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  2. 2016Communication-Efficient Learning of Deep Networks from Decentralized Datacited by 7,592
  3. 2018Split learning for health: Distributed deep learning without sharing raw patient datacited by 391
  4. 2020SplitFed: When Federated Learning Meets Split Learningcited by 308
  5. 2022Balanced Multimodal Learning via On-the-fly Gradient Modulationcited by 259
  6. 2019BottleNet++: An End-to-End Approach for Feature Compression in Device-Edge Co-Inference Systemscited by 94
  7. 2023Split Learning in 6G Edge Networkscited by 68
  8. 2024MBQ: Modality-Balanced Quantization for Large Vision-Language Modelscited by 45
  9. 2023Enhancing multimodal cooperation via sample-level modality valuationcited by 43
  10. 2022MM-SHAP: A Performance-agnostic Metric for Measuring Multimodal Contributions in Vision and Language Models & Taskscited by 42
  11. 2020A Federated Learning Framework for Healthcare IoT devicescited by 22
  12. 2023Reducing Communication for Split Learning by Randomized Top-k Sparsificationcited by 22
11 more
  1. 2023Communication-Efficient Split Learning via Adaptive Feature-Wise Compressioncited by 18
  2. 2025Improving Multimodal Learning via Imbalanced Learningcited by 16
  3. 2022SHAPE: An Unified Approach to Evaluate the Contribution and Cooperation of Individual Modalitiescited by 11
  4. 2024Communication-Efficient Multimodal Federated Learning: Joint Modality and Client Selectioncited by 10
  5. 2025Distributed LLMs and Multimodal Large Language Models: A Survey on Advances, Challenges, and Future Directionscited by 10
  6. 2022C3-SL: Circular Convolution-Based Batch-Wise Compression for Communication-Efficient Split Learningcited by 8
  7. 2023Split Federated Learning for 6G Enabled-Networks: Requirements, Challenges and Future Directionscited by 8
  8. 2023Adaptive Compression-Aware Split Learning and Inference for Enhanced Network Efficiencycited by 8
  9. 2025SL-ACC: A Communication-Efficient Split Learning Framework with Adaptive Channel-wise Compressioncited by 6
  10. 2025Contribution-Guided Asymmetric Learning for Robust Multimodal Fusion under Imbalance and Noisecited by 2
  11. 2026AutoEncoder-Compressed Parallel Split Learning for Pre-trained Model Fine-Tuningcited by 1

Built on by

Not cited yet.