Table 1.

The simple test results on the sensitivity of the neural network layer to pruning. The test neural network is LeNet-300-100, and which is a fully connected network. There are different pruning ratio of each layer which are shown in the first three columns, and the last column is the accuracy of the pruned networks.

Pruning Ratio of Layer1	Pruning Ratio of Layer2	Pruning Ratio of Layer3	Accuracy
0.8	0.8	0.8	0.6701
0.82	0.68	0.1	0.9386
0.84	0.5	0.4	0.9294
0.86	0.35	0.3	0.9050
0.88	0.2	0.2	0.8764
0.90	0.05	0	0.8399