Robust Image Classification via Using Multiple Diversity Losses

Yi Fang, Wen-Hao Zheng, Qihui Wang, Xiao-Xin Li
Proceedings of the 15th Asian Conference on Machine Learning, PMLR 222:359-373, 2024.

Abstract

Many research works focus on the robustness of convolutional neural networks (CNNs) on image classification. Diversity loss has been demonstrated to be an effective method to boost robustness. However, the existing diversity losses did not fully consider the strong correlation between regional features when filters are locally activated. They focused on improving filter responses constraint with classification loss. However, diversity loss has deeper optimization space. We explore the combinations of different filter diversity losses and feature diversity losses. We enhance the orthogonality between pair-wise filters to make them more diverse and penalize irrelevance between regional response mappings. We make multiple combinations and propose several methods on improving orthogonality, which have different adaptations for different datasets and network models. We evaluate their effectiveness in experiment. Our combinations could improve the efficiency of robust image recognition.

Cite this Paper

BibTeX
@InProceedings{pmlr-v222-fang24a, title = {Robust Image Classification via Using Multiple Diversity Losses}, author = {Fang, Yi and Zheng, Wen-Hao and Wang, Qihui and Li, Xiao-Xin}, booktitle = {Proceedings of the 15th Asian Conference on Machine Learning}, pages = {359--373}, year = {2024}, editor = {Yanıkoğlu, Berrin and Buntine, Wray}, volume = {222}, series = {Proceedings of Machine Learning Research}, month = {11--14 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v222/fang24a/fang24a.pdf}, url = {https://proceedings.mlr.press/v222/fang24a.html}, abstract = {Many research works focus on the robustness of convolutional neural networks (CNNs) on image classification. Diversity loss has been demonstrated to be an effective method to boost robustness. However, the existing diversity losses did not fully consider the strong correlation between regional features when filters are locally activated. They focused on improving filter responses constraint with classification loss. However, diversity loss has deeper optimization space. We explore the combinations of different filter diversity losses and feature diversity losses. We enhance the orthogonality between pair-wise filters to make them more diverse and penalize irrelevance between regional response mappings. We make multiple combinations and propose several methods on improving orthogonality, which have different adaptations for different datasets and network models. We evaluate their effectiveness in experiment. Our combinations could improve the efficiency of robust image recognition.} }
Endnote
%0 Conference Paper %T Robust Image Classification via Using Multiple Diversity Losses %A Yi Fang %A Wen-Hao Zheng %A Qihui Wang %A Xiao-Xin Li %B Proceedings of the 15th Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Berrin Yanıkoğlu %E Wray Buntine %F pmlr-v222-fang24a %I PMLR %P 359--373 %U https://proceedings.mlr.press/v222/fang24a.html %V 222 %X Many research works focus on the robustness of convolutional neural networks (CNNs) on image classification. Diversity loss has been demonstrated to be an effective method to boost robustness. However, the existing diversity losses did not fully consider the strong correlation between regional features when filters are locally activated. They focused on improving filter responses constraint with classification loss. However, diversity loss has deeper optimization space. We explore the combinations of different filter diversity losses and feature diversity losses. We enhance the orthogonality between pair-wise filters to make them more diverse and penalize irrelevance between regional response mappings. We make multiple combinations and propose several methods on improving orthogonality, which have different adaptations for different datasets and network models. We evaluate their effectiveness in experiment. Our combinations could improve the efficiency of robust image recognition.
APA
Fang, Y., Zheng, W., Wang, Q. & Li, X.. (2024). Robust Image Classification via Using Multiple Diversity Losses. Proceedings of the 15th Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 222:359-373 Available from https://proceedings.mlr.press/v222/fang24a.html.

Related Material