Squeeze and Excitation Networks

Squeeze and Excitation (SE) Module

The Squeeze and Excitation (SE) Module [1].

Performance

	Basic CNN	SENet (CNN + SE Module)
Accuracy	0.99340	0.99350
Precision	0.99339	0.99344
Recall	0.99329	0.99342
F1-Score	0.99334	0.99342

Basic CNN

Confusion Matrix
[[ 979    0    0    0    0    0    0    1    0    0]
 [   0 1132    0    1    0    0    1    1    0    0]
 [   0    0 1029    0    0    0    0    3    0    0]
 [   0    0    1 1006    0    3    0    0    0    0]
 [   0    0    1    0  975    0    2    0    0    4]
 [   1    0    0    7    0  882    1    0    0    1]
 [   4    2    0    0    0    1  950    0    1    0]
 [   1    3    3    2    0    0    0 1018    1    0]
 [   3    0    1    1    0    1    0    0  966    2]
 [   0    0    0    1    6    2    0    3    0  997]]
Class-0 | Precision: 0.99089, Recall: 0.99898, F1-Score: 0.99492
Class-1 | Precision: 0.99560, Recall: 0.99736, F1-Score: 0.99648
Class-2 | Precision: 0.99420, Recall: 0.99709, F1-Score: 0.99565
Class-3 | Precision: 0.98821, Recall: 0.99604, F1-Score: 0.99211
Class-4 | Precision: 0.99388, Recall: 0.99287, F1-Score: 0.99338
Class-5 | Precision: 0.99213, Recall: 0.98879, F1-Score: 0.99045
Class-6 | Precision: 0.99581, Recall: 0.99165, F1-Score: 0.99372
Class-7 | Precision: 0.99220, Recall: 0.99027, F1-Score: 0.99124
Class-8 | Precision: 0.99793, Recall: 0.99179, F1-Score: 0.99485
Class-9 | Precision: 0.99303, Recall: 0.98811, F1-Score: 0.99056

Total | Accuracy: 0.99340, Precision: 0.99339, Recall: 0.99329, F1-Score: 0.99334

SENet (Basic CNN + SE Module)

Confusion Matrix
[[ 978    1    0    0    0    0    0    1    0    0]
 [   0 1134    1    0    0    0    0    0    0    0]
 [   1    1 1024    0    0    0    0    5    0    1]
 [   0    0    0 1005    0    2    0    1    2    0]
 [   0    0    2    0  976    0    2    0    0    2]
 [   1    0    0    4    0  885    1    0    0    1]
 [   3    2    0    0    2    1  949    0    1    0]
 [   0    3    4    0    0    0    0 1020    0    1]
 [   2    0    1    0    1    1    0    0  966    3]
 [   0    0    0    0    4    2    0    3    0 1000]]
Class-0 | Precision: 0.99289, Recall: 0.99796, F1-Score: 0.99542
Class-1 | Precision: 0.99387, Recall: 0.99912, F1-Score: 0.99649
Class-2 | Precision: 0.99225, Recall: 0.99225, F1-Score: 0.99225
Class-3 | Precision: 0.99604, Recall: 0.99505, F1-Score: 0.99554
Class-4 | Precision: 0.99288, Recall: 0.99389, F1-Score: 0.99338
Class-5 | Precision: 0.99327, Recall: 0.99215, F1-Score: 0.99271
Class-6 | Precision: 0.99685, Recall: 0.99061, F1-Score: 0.99372
Class-7 | Precision: 0.99029, Recall: 0.99222, F1-Score: 0.99125
Class-8 | Precision: 0.99690, Recall: 0.99179, F1-Score: 0.99434
Class-9 | Precision: 0.99206, Recall: 0.99108, F1-Score: 0.99157

Total | Accuracy: 0.99370, Precision: 0.99373, Recall: 0.99361, F1-Score: 0.99367

Requirements

• Python 3.6.8
• Tensorflow 1.14.0
• Numpy 1.17.1
• Matplotlib 3.1.1

Reference

[1] Hu, Jie, et al. (2018). Squeeze-and-excitation networks. Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.