FeDMRA: Federated Incremental Learning with Dynamic Memory Replay Allocation

Tiantian Wang; Xiang Xiang; Simon S. Du

cs.LGMar 30, 2026

FeDMRA: Federated Incremental Learning with Dynamic Memory Replay Allocation

Authors: Tiantian Wang, Xiang Xiang, Simon S. Du

Organizations: Huazhong University of Science and Technology, Wuhan, China · University of Washington, Seattle, USA

Abstract

In federated healthcare systems, Federated Class-Incremental Learning (FCIL) has emerged as a key paradigm, enabling continuous adaptive model learning among distributed clients while safeguarding data privacy. However, in practical applications, data across agent nodes within the distributed framework often exhibits non-independent and identically distributed (non-IID) characteristics, rendering traditional continual learning methods inapplicable. To address these challenges, this paper covers more comprehensive incremental task scenarios and proposes a dynamic memory allocation strategy for exemplar storage based on the data replay mechanism. This strategy fully taps into the inherent potential of data heterogeneity, while taking into account the performance fairness of all participating clients, thereby establishing a balanced and adaptive solution to mitigate catastrophic forgetting. Unlike the fixed allocation of client exemplar memory, the proposed scheme emphasizes the rational allocation of limited storage resources among clients to improve model performance. Furthermore, extensive experiments are conducted on three medical image datasets, and the results demonstrate significant performance improvements compared to existing baseline models.

Figures & tables

Figure 1: The legend reflects our motivations and contributions. The coordinate system is used to characterize the sample distribution of the client. Compared to common replay-based methods that allocate fixed memory size ${m_{fix}}$ per client and uniformly distribute it across all classes, FeDMRA leverages the heterogeneity of client data distributions and their performance contributions to dynamically allocate matched memory sizes ${m_{c}}$ and ${m_{c}^{y}}$ , thereby optimizing the influence of stored exemplars.

Figure 2: An overview of the proposed FeDMRA. Server : 1. Receive the model parameters $w_{c}$ and class distribution uploaded by the clients. 2. Calculate the storage size of example sets for each category on the client side through client-level and class-level calculations. 3. Aggregate the client models using FedAvg to update the global model. Client : 1. Filter samples to fill the allocated size ${m_{c}^{y}}$ of the example set and replay. 2. Use the global model $w_{{g,t}}$ as a teacher model to perform knowledge transfer for the local model. 3. Local incremental training updates the model and uploads it.

Method	FCIL-M		FCIL-A		FCIL-H		FDIL		FCDIL
Method	$\mathcal{A}_{avg}\uparrow$	$\mathcal{A}_{last}\uparrow$	$\mathcal{A}_{avg}\uparrow$	$\mathcal{A}_{last}\uparrow$	$\mathcal{A}_{avg}\uparrow$	$\mathcal{A}_{last}\uparrow$	$\mathcal{A}_{avg}\uparrow$	$\mathcal{A}_{last}\uparrow$	$\mathcal{A}_{avg}\uparrow$	$\mathcal{A}_{last}\uparrow$
UP	86.61	85.5	88.62	88.78	95.06	95.87	84.10	89.18	87.30	88.45
iCaRL+FL	72.05	58.88	65.36	57.3	61.89	46.51	64.57	70.96	55.41	55.56
UACL+FL	52.06	25.20	54.62	49.32	66.59	48.44	51.21	36.53	51.93	52.52
MFCL	69.75	20.87	46.00	31.91	79.27	51.09	74.02	64.82	51.12	29.16
DDDR	68.14	64.34	51.38	30.42	78.23	64.00	77.56	69.05	60.34	36.50

Table 1: Report the classification accuracy

A_{last}

and

A_{avg}

under heterogeneity

\alpha=1.0

(%).

Method	FCIL-M		FCIL-A		FCIL-H		FDIL		FCDIL
Method	$\mathcal{A}_{avg}\uparrow$	$\mathcal{A}_{last}\uparrow$	$\mathcal{A}_{avg}\uparrow$	$\mathcal{A}_{last}\uparrow$	$\mathcal{A}_{avg}\uparrow$	$\mathcal{A}_{last}\uparrow$	$\mathcal{A}_{avg}\uparrow$	$\mathcal{A}_{last}\uparrow$	$\mathcal{A}_{avg}\uparrow$	$\mathcal{A}_{last}\uparrow$
UP	86.61	85.50	88.62	88.78	95.06	95.87	84.10	89.18	83.61	86.95
iCaRL+FL	72.05	58.88	65.36	57.30	61.89	46.51	64.57	70.96	80.53	77.49
UACL+FL	52.06	25.20	54.62	49.32	66.59	48.44	51.21	46.05	51.60	53.63
MFCL	56.40	15.34	39.25	8.37	77.91	51.20	75.50	68.48	38.92	18.86
DDDR	40.75	17.80	48.74	28.15	73.21	59.02	76.01	72.30	41.17	14.72

Table 2: Report the classification accuracy

A_{last}

and

A_{avg}

under heterogeneity

\alpha=0.5

(%).

$\bm{\alpha}$	$0.5$				$1.0$				$5.0$
$\bm{a}$	$0.8$		$0.4$		$0.8$		$0.4$		$0.8$		$0.4$
$\bm{\lambda}$	$0.8$	$0.4$	$0.8$	$0.4$	$0.8$	$0.4$	$0.8$	$0.4$	$0.8$	$0.4$	$0.8$	$0.4$
FCIL-M	76.19	77.08	76.91	77.29	82.12	82.58	82.44	83.06	80.97	80.87	82.66	82.90
FCIL-A	78.45	79.36	80.00	78.36	82.75	82.27	82.91	83.26	84.23	83.80	83.77	83.89
FCIL-H	87.83	88.48	88.26	88.07	91.51	91.16	91.77	91.72	91.71	93.19	91.03	91.75
FDIL	72.44	73.13	75.83	72.41	80.70	80.72	82.71	81.55	80.75	81.05	81.97	81.36

Table 3: Average accuracy with different configurations under various heterogeneities

\alpha

(%).

FCIL-H

FDIL

FCDIL

dy

\checkmark

\checkmark

\times

\times

\checkmark

\checkmark

\times

\times

\checkmark

\checkmark

\times

\times

\mathcal{L}_{MG}

\checkmark

\times

\checkmark

\times

\checkmark

\times

\checkmark

\times

\checkmark

\times

\checkmark

\times

\mathcal{L}_{KL}

\checkmark

\times

\times

\checkmark

\checkmark

\times

\times

\checkmark

\checkmark

\times

\times

\checkmark

A_{avg}\uparrow

91.77

91.06

90.98

90.51

82.71

81.60

78.61

81.82

85.26

82.74

78.98

80.98

A_{last}\uparrow

87.34

87.58

84.30

85.90

87.64

87.70

86.51

88.20

83.35

82.43

61.41

71.52

Table 4: Ablation study of FeDMRA (%).

Panel of Figure 3 — Figure 3: Our method under different configuration (a) weight $\delta$ of $\mathcal{L}_{MG}$ , (b) $m_{max}$ for FCIL, (c) $m_{max}$ for FDIL and FCDIL.

Appendix figures & tables2 assets

Supplementary material from the paper’s appendix.

Appendix

Task	Class Number per Task
FCIL-M	4, 3, 3, 3
FCIL-A	4, 3, 3
FCIL-H	4, 2, 2
FDIL	13, 10, 8
FCDIL	4, 3, 3, 3, 4, 3, 3, 4, 4

Appendix

Table 5: Task settings of FCIL, FDIL, and FCDIL

Method	Memory Size of Each Task
iCaRL+FL (Per-Client)	240 $\times$ 5	240 $\times$ 6	240 $\times$ 6	240 $\times$ 7
UACL+FL (Per-Client)	240 $\times$ 5	240 $\times$ 6	240 $\times$ 6	240 $\times$ 7
MFCL (Per-Batch)	128 $\times$ BS	128 $\times$ BS	128 $\times$ BS	128 $\times$ BS
DDDR (Per-Class)	240 $\times$ $Y^{t}$	240 $\times$ $Y^{t}$	240 $\times$ $Y^{t}$	240 $\times$ $Y^{t}$
ReFed (Per-Client)	240 $\times$ 5	240 $\times$ 6	240 $\times$ 6	240 $\times$ 7
FeDMRA (Per-Client)	240 $\times$ 5	322,322,190,209,152,240	311,400,47,309,132,190	346,61,367,122,370,134,240

Appendix

Table 6: Report the cache size of example-set for each method. Taking FCIL-M as an example.

Federated Continual Learning

Federated continual learning enables distributed clients to adapt models to changing data while retaining past knowledge and keeping raw data local.

See all 35 papers

From Selection to Scheduling: Federated Geometry-Aware Correction Makes Exemplar Replay Work Better under Continual Dynamic Heterogeneity
Apr 9, 2026Zhuang Qi, Ying-Peng Tang, Lei Meng +4Federated Continual Learning Replay-Based Continual Learning
School of Software, Shandong University, China · College of Computing and Data Science, Nanyang Technological University, Singapore · School of Computer Science and Technology, Harbin Institute of Technology, China
Task-Agnostic Federated Continual Learning via Replay-Free Gradient Projection
Sep 25, 2025Seohyeon Cha, Huancheng Chen, Haris VikaloFederated Continual Learning Gradient Projection
Department of Electrical and Computer Engineering The University of Texas at Austin
Federated Continual Learning as a Distributed Drift-Plus-Penalty Control Problem
Aug 21, 2026Nazreen Shah, Naveen Kumar Reddy Somireddy, Zubair Shaban +2Federated Continual Learning Catastrophic Forgetting
IIIT Delhi India · IIT Dharwad India

Apr 9, 2026J/K move · Enter open · S save

From Selection to Scheduling: Federated Geometry-Aware Correction Makes Exemplar Replay Work Better under Continual Dynamic Heterogeneity

Zhuang Qi, Ying-Peng Tang, Lei Meng, Guoqing Chao+3

School of Software, Shandong University, China · College of Computing and Data Science, Nanyang Technological University, Singapore · School of Computer Science and Technology, Harbin Institute of Technology, China

Federated Continual Learning Replay-Based Continual Learning Heterogeneous Federated Learning Class-Imbalanced Learning

Open paper AskPDF arXiv

Exemplar replay has become an effective strategy for mitigating catastrophic forgetting in federated continual learning (FCL) by retaining representative samples from past tasks. Existing studies focus on designing sample-importance estimation mechanisms to identify information-rich samples. However, they typically overlook strategies for effectively utilizing the selected exemplars, which limits their performance under continual dynamic heterogeneity across clients and tasks. To address this issue, this paper proposes a Federated gEometry-Aware correcTion method, termed FEAT, which alleviates imbalance-induced representation collapse that drags rare-class features toward frequent classes across clients. Specifically, it consists of two key modules: 1) the Geometric Structure Alignment module performs structural knowledge distillation by aligning the pairwise angular similarities between feature representations and their corresponding Equiangular Tight Frame prototypes, which are fixed and shared across clients to serve as a class-discriminative reference structure. This encourages geometric consistency across tasks and helps mitigate representation drift; 2) the Energy-based Geometric Correction module removes task-irrelevant directional components from feature embeddings, which reduces prediction bias toward majority classes. This improves sensitivity to minority classes and enhances the model's robustness under class-imbalanced distributions.

Abstract

Federated Continual Learning

Explore similar work

From Selection to Scheduling: Federated Geometry-Aware Correction Makes Exemplar Replay Work Better under Continual Dynamic Heterogeneity

Task-Agnostic Federated Continual Learning via Replay-Free Gradient Projection

Federated Continual Learning as a Distributed Drift-Plus-Penalty Control Problem