Dynamic Allocation Hypernetwork with Adaptive Model Recalibration for FCL

2025-03-23

AI Summary

To address catastrophic forgetting and optimization bias induced by asynchronous tasks in medical federated continual learning (FCL), this paper proposes DAHyper-AMR—a novel framework comprising Dynamic Allocation Hypernetwork (DAHyper) and Adaptive Model Recalibration (AMR). DAHyper dynamically generates task-specific parameters guided by task identifiers, enabling fine-grained adaptation to heterogeneous, evolving client task streams. AMR mitigates knowledge loss by temporally weighted fusion of historical model states, leveraging chronological similarity to preserve prior task knowledge. Together, these components jointly alleviate cross-client knowledge forgetting and objective conflicts under dynamic, non-stationary task distributions. Evaluated on the multi-center AMOS dataset, DAHyper-AMR significantly outperforms existing FCL methods, achieving superior model generalization and sustained long-term performance without compromising privacy or requiring centralized data access.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Multiagent Systems: Multiagent Learning
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Federated continual learning (FCL) offers an emerging pattern to facilitate the applicability of federated learning (FL) in real-world scenarios, where tasks evolve dynamically and asynchronously across clients, especially in medical scenario. Existing server-side FCL methods in nature domain construct a continually learnable server model by client aggregation on all-involved tasks. However, they are challenged by: (1) Catastrophic forgetting for previously learned tasks, leading to error accumulation in server model, making it difficult to sustain comprehensive knowledge across all tasks. (2) Biased optimization due to asynchronous tasks handled across different clients, leading to the collision of optimization targets of different clients at the same time steps. In this work, we take the first step to propose a novel server-side FCL pattern in medical domain, Dynamic Allocation Hypernetwork with adaptive model recalibration ( extbf{FedDAH}). It is to facilitate collaborative learning under the distinct and dynamic task streams across clients. To alleviate the catastrophic forgetting, we propose a dynamic allocation hypernetwork (DAHyper) where a continually updated hypernetwork is designed to manage the mapping between task identities and their associated model parameters, enabling the dynamic allocation of the model across clients. For the biased optimization, we introduce a novel adaptive model recalibration (AMR) to incorporate the candidate changes of historical models into current server updates, and assign weights to identical tasks across different time steps based on the similarity for continual optimization. Extensive experiments on the AMOS dataset demonstrate the superiority of our FedDAH to other FCL methods on sites with different task streams. The code is available:https://github.com/jinlab-imvr/FedDAH.

Problem

• Address catastrophic forgetting in federated continual learning
• Mitigate biased optimization from asynchronous client tasks
• Enable dynamic model allocation across evolving medical tasks

Innovation

• Dynamic Allocation Hypernetwork for task mapping
• Adaptive Model Recalibration for biased optimization
• Continually updated hypernetwork prevents catastrophic forgetting