Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Fig 1.

The overview of system model.

More »

Fig 1 Expand

Table 1.

Comparison with conventional hyperparameter search.

More »

Table 1 Expand

Table 2.

Summary of experimental datasets and partitioning strategies.

More »

Table 2 Expand

Table 3.

Experimental setup.

More »

Table 3 Expand

Table 4.

Hyperparameter settings for all compared methods.

More »

Table 4 Expand

Fig 2.

IID settings for different datasets.

More »

Fig 2 Expand

Fig 3.

Non-IID settings for different datasets.

More »

Fig 3 Expand

Table 5.

Function comparison of different schemes.

More »

Table 5 Expand

Fig 4.

Comparison of different schemes (CIFAR-10 dataset).

More »

Fig 4 Expand

Fig 5.

Comparison of different schemes (MNIST dataset).

More »

Fig 5 Expand

Table 6.

Sensitivity analysis of gradient sparsification ratio k on CIFAR-10, VGG11, 20 rounds.

More »

Table 6 Expand

Fig 6.

(Edge device computing overhead) comparison of different schemes.

More »

Fig 6 Expand

Fig 7.

(Cloud server computing cost) comparison of different schemes.

More »

Fig 7 Expand

Table 7.

Trade-off between decision cost and communication benefit.

More »

Table 7 Expand

Table 8.

System-level profiling — CIFAR-10 / VGG11 / C = 10 / 10 Mbps uplink.

More »

Table 8 Expand

Table 9.

System-level profiling — MNIST / 2NN / C = 10 / 10 Mbps uplink.

More »

Table 9 Expand

Table 10.

Ablation study results — CIFAR-10 dataset (C = 10, 20 rounds, VGG11).

More »

Table 10 Expand

Table 11.

Ablation study results — MNIST dataset (C = 10, 20 rounds, 2NN).

More »

Table 11 Expand

Table 12.

Comparison of fixed-rate pruning and APMKFL on CIFAR-10 and MNIST datasets.

More »

Table 12 Expand