TY - JOUR
T1 - OSLNet
T2 - Deep Small-Sample Classification with an Orthogonal Softmax Layer
AU - Li, Xiaoxu
AU - Chang, Dongliang
AU - Ma, Zhanyu
AU - Tan, Zheng-Hua
AU - Xue, Jing-Hao
AU - Cao, Jie
AU - Yu, Jingyi
AU - Guo, Jun
PY - 2020
Y1 - 2020
N2 - A deep neural network of multiple nonlinear layers forms a large function space, which can easily lead to overfitting when it encounters small-sample data. To mitigate overfitting in small-sample classification, learning more discriminative features from small-sample data is becoming a new trend. To this end, this paper aims to find a subspace of neural networks that can facilitate a large decision margin. Specifically, we propose the Orthogonal Softmax Layer (OSL), which makes the weight vectors in the classification layer remain orthogonal during both the training and test processes. The Rademacher complexity of a network using the OSL is only 1 K, where K is the number of classes, of that of a network using the fully connected classification layer, leading to a tighter generalization error bound. Experimental results demonstrate that the proposed OSL has better performance than the methods used for comparison on four small-sample benchmark datasets, as well as its applicability to large-sample datasets. Codes are available at: https://github.com/dongliangchang/OSLNet.
AB - A deep neural network of multiple nonlinear layers forms a large function space, which can easily lead to overfitting when it encounters small-sample data. To mitigate overfitting in small-sample classification, learning more discriminative features from small-sample data is becoming a new trend. To this end, this paper aims to find a subspace of neural networks that can facilitate a large decision margin. Specifically, we propose the Orthogonal Softmax Layer (OSL), which makes the weight vectors in the classification layer remain orthogonal during both the training and test processes. The Rademacher complexity of a network using the OSL is only 1 K, where K is the number of classes, of that of a network using the fully connected classification layer, leading to a tighter generalization error bound. Experimental results demonstrate that the proposed OSL has better performance than the methods used for comparison on four small-sample benchmark datasets, as well as its applicability to large-sample datasets. Codes are available at: https://github.com/dongliangchang/OSLNet.
KW - Deep neural network
KW - Orthogonal softmax layer
KW - overfitting
KW - small-sample classification
UR - http://www.scopus.com/inward/record.url?scp=85087547443&partnerID=8YFLogxK
U2 - 10.1109/TIP.2020.2990277
DO - 10.1109/TIP.2020.2990277
M3 - Journal article
SN - 1057-7149
VL - 29
SP - 6482
EP - 6495
JO - I E E E Transactions on Image Processing
JF - I E E E Transactions on Image Processing
M1 - 9088302
ER -