Ada$^2$NPT: An Adaptive Nearest Proxies Triplet Loss for Attribute-Aware Face Recognition with Adaptively Compacted Feature Learning

Lei Ju, Zhanhua Feng, Muhammad Awais, Josef Kittler
Proceedings of the 15th Asian Conference on Machine Learning, PMLR 222:614-629, 2024.

Abstract

Attribute-aware face recognition has gained increasing attention in recent years due to its potential to improve the robustness of face recognition systems. However, this may raise concerns about potential biases and privacy issues. To alleviate this, some studies involve complex designs to obtain independent ID and attribute features and fuse them based on the application scenario (for better accuracy or fairness). In this paper, we obviate their complex design and demonstrate that the Nearest neighbours Proxy Triplet (NPT) loss has an intrinsic capability for feature disentanglement. To further enhance the effectiveness of NPT, we propose a novel margin-based loss, namely Adaptive-rank NPT, which naturally separates the identity and attribute features. While a margin-based loss ensures inter-class separability, it imposes no constraints on intra-class compactness. The samples that meet the inter-class margin will not contribute to network training. To mitigate this issue, we propose an adaptive distance measurement to promote the compactness of the learned features, resulting in the final Ada$^2$NPT loss. The experimental results obtained on several benchmarks demonstrate the superiority and merits of the proposed loss function over the state-of-the-art losses in terms of accuracy and fairness.

Cite this Paper

BibTeX
@InProceedings{pmlr-v222-ju24a, title = {{Ada$^2$NPT}: {A}n Adaptive Nearest Proxies Triplet Loss for Attribute-Aware Face Recognition with Adaptively Compacted Feature Learning}, author = {Ju, Lei and Feng, Zhanhua and Awais, Muhammad and Kittler, Josef}, booktitle = {Proceedings of the 15th Asian Conference on Machine Learning}, pages = {614--629}, 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/ju24a/ju24a.pdf}, url = {https://proceedings.mlr.press/v222/ju24a.html}, abstract = {Attribute-aware face recognition has gained increasing attention in recent years due to its potential to improve the robustness of face recognition systems. However, this may raise concerns about potential biases and privacy issues. To alleviate this, some studies involve complex designs to obtain independent ID and attribute features and fuse them based on the application scenario (for better accuracy or fairness). In this paper, we obviate their complex design and demonstrate that the Nearest neighbours Proxy Triplet (NPT) loss has an intrinsic capability for feature disentanglement. To further enhance the effectiveness of NPT, we propose a novel margin-based loss, namely Adaptive-rank NPT, which naturally separates the identity and attribute features. While a margin-based loss ensures inter-class separability, it imposes no constraints on intra-class compactness. The samples that meet the inter-class margin will not contribute to network training. To mitigate this issue, we propose an adaptive distance measurement to promote the compactness of the learned features, resulting in the final Ada$^2$NPT loss. The experimental results obtained on several benchmarks demonstrate the superiority and merits of the proposed loss function over the state-of-the-art losses in terms of accuracy and fairness.} }
Endnote
%0 Conference Paper %T Ada$^2$NPT: An Adaptive Nearest Proxies Triplet Loss for Attribute-Aware Face Recognition with Adaptively Compacted Feature Learning %A Lei Ju %A Zhanhua Feng %A Muhammad Awais %A Josef Kittler %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-ju24a %I PMLR %P 614--629 %U https://proceedings.mlr.press/v222/ju24a.html %V 222 %X Attribute-aware face recognition has gained increasing attention in recent years due to its potential to improve the robustness of face recognition systems. However, this may raise concerns about potential biases and privacy issues. To alleviate this, some studies involve complex designs to obtain independent ID and attribute features and fuse them based on the application scenario (for better accuracy or fairness). In this paper, we obviate their complex design and demonstrate that the Nearest neighbours Proxy Triplet (NPT) loss has an intrinsic capability for feature disentanglement. To further enhance the effectiveness of NPT, we propose a novel margin-based loss, namely Adaptive-rank NPT, which naturally separates the identity and attribute features. While a margin-based loss ensures inter-class separability, it imposes no constraints on intra-class compactness. The samples that meet the inter-class margin will not contribute to network training. To mitigate this issue, we propose an adaptive distance measurement to promote the compactness of the learned features, resulting in the final Ada$^2$NPT loss. The experimental results obtained on several benchmarks demonstrate the superiority and merits of the proposed loss function over the state-of-the-art losses in terms of accuracy and fairness.
APA
Ju, L., Feng, Z., Awais, M. & Kittler, J.. (2024). Ada$^2$NPT: An Adaptive Nearest Proxies Triplet Loss for Attribute-Aware Face Recognition with Adaptively Compacted Feature Learning. Proceedings of the 15th Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 222:614-629 Available from https://proceedings.mlr.press/v222/ju24a.html.

Related Material