Federated Class-Incremental Learning

Benchmarks

Dataset Name	SOTA Method	Metric
Tiny-ImageNet 10 tasks (20 classes per task) (test)		FAA84.3	54	3mo ago
CIFAR-100 Quantity-based label imbalance		FAA88.6	42	3mo ago
CIFAR-100 Distribution-based label imbalance		FAA90.1	39	3mo ago
VTAB	FedRAN	Average Accuracy92.49	21	1mo ago
CIFAR-100	FedRAN	Average FAA76.66	21	1mo ago
CIFAR-100 Homogeneous (incremental-stream)	PRO-MAX	FAA46.72	19	1mo ago
TinyImageNet (test)	GLFC	Score_12049	17	4mo ago
Cora	PRO-MAX	Final Accuracy (FAA)70.42	14	1mo ago
FCIL-H		Average Accuracy (Aavg)95.06	14	3mo ago
FCIL-A		Aavg88.62	14	3mo ago
FCIL-M		Avg Acc86.61	14	3mo ago
ImageNet-R	C²Prompt	FAA (β=0.5)75.23	13	4mo ago
THUCNews 10	PRO-MAX	FAA Score85.16	12	1mo ago
CLINC150	PRO-MAX	FAA Score72.36	12	1mo ago
CIFAR-100 Heterogeneous (test)	FedAvg	Communication Overhead Factor1	11	1mo ago
TinyImageNet scale-up (Heterogeneous)	PRO-MAX	FAA37.54	11	1mo ago
TinyImageNet scale-up (Homogeneous)	PRO-MAX	FAA40.36	11	1mo ago
ImageNet Subset 10 incremental tasks	GLFC (Ours)	Accuracy42.7	10	4mo ago
TinyImageNet first 10 tasks 35	DDE + iCaRL + FL	Performance @ 10%70	10	4mo ago
FCIL-H alpha=0.5 (test)		Average Accuracy95.06	7	3mo ago
FCIL-A alpha=0.5 (test)		Average Accuracy (Aavg)88.62	7	3mo ago
FCIL-M alpha=0.5 (test)		Average Accuracy86.61	7	3mo ago
CIFAR-100 alpha = 6	Fed-TaLoRA	FAA76.6	5	3mo ago

Showing 23 of 23 rows