Out-of-Distribution Detection via Deep Multi-Comprehension Ensemble

Chenhui Xu, Fuxun Yu, Zirui Xu, Nathan Inkawhich, Xiang Chen
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:55465-55489, 2024.

Abstract

Recent research works demonstrate that one of the significant factors for the model Out-of-Distirbution detection performance is the scale of the OOD feature representation field. Consequently, model ensemble emerges as a trending method to expand this feature representation field leveraging expected model diversity. However, by proposing novel qualitative and quantitative model ensemble evaluation methods (i.e., Loss Basin/Barrier Visualization and Self-Coupling Index), we reveal that the previous ensemble methods incorporate affine-transformable weights with limited variability and fail to provide desired feature representation diversity. Therefore, we escalate the traditional model ensemble dimensions (different weight initialization, data holdout, etc.) into distinct supervision tasks, which we name as Multi-Comprehension (MC) Ensemble. MC Ensemble leverages various training tasks to form different comprehensions of the data and labels, resulting in the extension of the feature representation field. In experiments, we demonstrate the superior performance of the MC Ensemble strategy in the OOD detection task compared to both the naive Deep Ensemble method and the standalone model of comparable size.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-xu24ae, title = {Out-of-Distribution Detection via Deep Multi-Comprehension Ensemble}, author = {Xu, Chenhui and Yu, Fuxun and Xu, Zirui and Inkawhich, Nathan and Chen, Xiang}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {55465--55489}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/xu24ae/xu24ae.pdf}, url = {https://proceedings.mlr.press/v235/xu24ae.html}, abstract = {Recent research works demonstrate that one of the significant factors for the model Out-of-Distirbution detection performance is the scale of the OOD feature representation field. Consequently, model ensemble emerges as a trending method to expand this feature representation field leveraging expected model diversity. However, by proposing novel qualitative and quantitative model ensemble evaluation methods (i.e., Loss Basin/Barrier Visualization and Self-Coupling Index), we reveal that the previous ensemble methods incorporate affine-transformable weights with limited variability and fail to provide desired feature representation diversity. Therefore, we escalate the traditional model ensemble dimensions (different weight initialization, data holdout, etc.) into distinct supervision tasks, which we name as Multi-Comprehension (MC) Ensemble. MC Ensemble leverages various training tasks to form different comprehensions of the data and labels, resulting in the extension of the feature representation field. In experiments, we demonstrate the superior performance of the MC Ensemble strategy in the OOD detection task compared to both the naive Deep Ensemble method and the standalone model of comparable size.} }
Endnote
%0 Conference Paper %T Out-of-Distribution Detection via Deep Multi-Comprehension Ensemble %A Chenhui Xu %A Fuxun Yu %A Zirui Xu %A Nathan Inkawhich %A Xiang Chen %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-xu24ae %I PMLR %P 55465--55489 %U https://proceedings.mlr.press/v235/xu24ae.html %V 235 %X Recent research works demonstrate that one of the significant factors for the model Out-of-Distirbution detection performance is the scale of the OOD feature representation field. Consequently, model ensemble emerges as a trending method to expand this feature representation field leveraging expected model diversity. However, by proposing novel qualitative and quantitative model ensemble evaluation methods (i.e., Loss Basin/Barrier Visualization and Self-Coupling Index), we reveal that the previous ensemble methods incorporate affine-transformable weights with limited variability and fail to provide desired feature representation diversity. Therefore, we escalate the traditional model ensemble dimensions (different weight initialization, data holdout, etc.) into distinct supervision tasks, which we name as Multi-Comprehension (MC) Ensemble. MC Ensemble leverages various training tasks to form different comprehensions of the data and labels, resulting in the extension of the feature representation field. In experiments, we demonstrate the superior performance of the MC Ensemble strategy in the OOD detection task compared to both the naive Deep Ensemble method and the standalone model of comparable size.
APA
Xu, C., Yu, F., Xu, Z., Inkawhich, N. & Chen, X.. (2024). Out-of-Distribution Detection via Deep Multi-Comprehension Ensemble. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:55465-55489 Available from https://proceedings.mlr.press/v235/xu24ae.html.

Related Material