Table 5 Average accuracy (%) and training rounds to achieve target accuracy for personalized models.

From: An efficient personalized federated learning approach in heterogeneous environments: a reinforcement learning perspective

Dataset

FedAvg

Ditto

FedRep

FedPRL

Accuracy

Rounds

Accuracy

Rounds

Accuracy

Rounds

Accuracy

Rounds

CIFAR-10

71.9

159

81.5

92

82.3

119

87.5

64

CIFAR-100

60.4

163

68.0

112

69.1

135

72.9

72

FEMNIST

65.6

140

71.8

105

72.5

124

74.2

77

MobiAct (v1)

80.8

214

83.0

188

84.6

176

94.8

79

MobiAct (v2)

79.3

236

81.6

192

83.8

201

94.3

72

MobiAct (v3)

76.0

189

78.4

153

78.6

145

83.3

83

  1. Significant values in Bold.
Back to article page