Table 4 Comparison test of different methods.

From: ALDP-FL for adaptive local differential privacy in federated learning

Dataset

Model

Approach

Accuracy

Precision

Recall

F1

MNIST

LeNet-5

FedAvg

0.9961

0.9984

0.9957

0.9970

Fed-DPA

0.8735

0.8741

0.8745

0.8743

GFL-ALDPA

0.9218

0.9250

0.9226

0.9238

LDP-Fed

0.9150

0.9138

0.9144

0.9141

ALDP-FL

0.9725

0.9701

0.9750

0.9725

ResNet18

FedAvg

0.9996

0.9997

0.9986

0.9991

Fed-DPA

0.9213

0.9230

0.9237

0.9233

GFL-ALDPA

0.9512

0.9535

0.9536

0.9535

LDP-Fed

0.9531

0.9538

0.9547

0.9542

ALDP-FL

0.9823

0.9818

0.9781

0.9799

Fashion MNIST

LeNet-5

FedAvg

0.9135

0.9146

0.9134

0.9140

Fed-DPA

0.7327

0.7318

0.7340

0.7329

GFL-ALDPA

0.8341

0.8343

0.8327

0.8335

LDP-Fed

0.8120

0.8115

0.8147

0.8131

ALDP-FL

0.8528

0.8519

0.8524

0.8521

ResNet18

FedAvg

0.9542

0.9537

0.9523

0.9530

Fed-DPA

0.8148

0.8099

0.8139

0.8119

GFL-ALDPA

0.8835

0.8832

0.8831

0.8831

LDP-Fed

0.8640

0.8649

0.8627

0.8638

ALDP-FL

0.9129

0.9123

0.9092

0.9107

CIFAR-10

LeNet-5

FedAvg

0.6931

0.6920

0.6946

0.6933

Fed-DPA

0.4841

0.4837

0.4829

0.4833

GFL-ALDPA

0.5738

0.5728

0.5725

0.5726

LDP-Fed

0.5449

0.5438

0.5441

0.5439

ALDP-FL

0.6132

0.6137

0.6171

0.6154

ResNet18

FedAvg

0.9146

0.9136

0.9141

0.9138

Fed-DPA

0.7241

0.7245

0.7233

0.7239

GFL-ALDPA

0.8040

0.8037

0.8028

0.8032

LDP-Fed

0.7533

0.7541

0.7534

0.7537

ALDP-FL

0.8526

0.8476

0.8518

0.8497

Back to article page