TL;DR
Experiments show that FedAMB improves multimodal accuracy and missing-modality robustness and prevents the propagation of fusion-biased teachers while directly improving unimodal representations.
FedTaste is proposed, a parameter-efficient framework for topology-aware structural transfer in Multimodal Federated Learning with missing modalities that avoids explicit modality imputation while preserving shared semantic structure across clients.
The metrics of individual modality contribution (IMC) and multimodal synergistic gain (MSG) are introduced to quantify sample-level and semantic-level utility, so as to guide semantic denoising selection and robust conditional balancing strategies, effectively mitigating noise interference.
Flux is proposed, a multimodal federated learning framework built around two complementary components, modality-aware confidence tempering and gradient-decoupled private adaptation, that enables sample-specific, client-local confidence adaptation without allowing confidence-dependent gradients to perturb shared representation learning.
FedADB, a Class Anchor-Driven Dual-Branch FL framework, a dual-branch collaborative training mechanism designed for clients that achieves significant improvements in both accuracy and convergence speed.
Experiments show that LAUA substantially mitigates performance degradation under modality missingness across retrieval and regression tasks, attaining up to 20% relative improvement in MRR for retrieval and up to 24.3% relative improvement in MSE for regression.
Federated Learning (FL) has demonstrated a promising future in privacy-friendly collaboration but it faces the data heterogeneity problem. Knowledge Distillation (KD) can serve as an effective method to address this issue. However, challenges arise from the unreliability of existing distillation methods in multi-domain scenarios. Prevalent distillation solutions primarily aim to fit the distributions of the global model directly by minimizing forward Kullback-Leibler divergence (KLD). This results in significant bias when the outputs of the global model are multi-peaked, which indicates the unreliability of distillation pathway . Mean-while, cross-domain update conflicts can notably reduce the accuracy of the global model (teacher model) in certain domains, reflecting the unreliability of the teacher model in these domains. In this work, we propose DKDR ( D ynamic K nowledge D istillation for R eliability in Federated Learning), which dynamically assigns weights to forward and reverse KLD based on knowledge discrepancies. This enables clients to fit the outputs from the teacher precisely. Moreover, we use knowledge decoupling to identify domain experts, thus clients can acquire reliable domain knowledge from experts. Empirical results from single-domain and multi-domain image classification tasks demonstrate the effectiveness of the proposed method and the efficiency of its key modules. The code is available at https://github.com/YueyangYuan/DKDR .