Predictive Dynamic Fusion

Bing Cao, Yinan Xia, Yi Ding, Changqing Zhang, Qinghua Hu
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:5608-5628, 2024.

Abstract

Multimodal fusion is crucial in joint decision-making systems for rendering holistic judgments. Since multimodal data changes in open environments, dynamic fusion has emerged and achieved remarkable progress in numerous applications. However, most existing dynamic multimodal fusion methods lack theoretical guarantees and easily fall into suboptimal problems, yielding unreliability and instability. To address this issue, we propose a Predictive Dynamic Fusion (PDF) framework for multimodal learning. We proceed to reveal the multimodal fusion from a generalization perspective and theoretically derive the predictable Collaborative Belief (Co-Belief) with Mono- and Holo-Confidence, which provably reduces the upper bound of generalization error. Accordingly, we further propose a relative calibration strategy to calibrate the predicted Co-Belief for potential uncertainty. Extensive experiments on multiple benchmarks confirm our superiority. Our code is available at https://github.com/Yinan-Xia/PDF.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-cao24c, title = {Predictive Dynamic Fusion}, author = {Cao, Bing and Xia, Yinan and Ding, Yi and Zhang, Changqing and Hu, Qinghua}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {5608--5628}, 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/cao24c/cao24c.pdf}, url = {https://proceedings.mlr.press/v235/cao24c.html}, abstract = {Multimodal fusion is crucial in joint decision-making systems for rendering holistic judgments. Since multimodal data changes in open environments, dynamic fusion has emerged and achieved remarkable progress in numerous applications. However, most existing dynamic multimodal fusion methods lack theoretical guarantees and easily fall into suboptimal problems, yielding unreliability and instability. To address this issue, we propose a Predictive Dynamic Fusion (PDF) framework for multimodal learning. We proceed to reveal the multimodal fusion from a generalization perspective and theoretically derive the predictable Collaborative Belief (Co-Belief) with Mono- and Holo-Confidence, which provably reduces the upper bound of generalization error. Accordingly, we further propose a relative calibration strategy to calibrate the predicted Co-Belief for potential uncertainty. Extensive experiments on multiple benchmarks confirm our superiority. Our code is available at https://github.com/Yinan-Xia/PDF.} }
Endnote
%0 Conference Paper %T Predictive Dynamic Fusion %A Bing Cao %A Yinan Xia %A Yi Ding %A Changqing Zhang %A Qinghua Hu %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-cao24c %I PMLR %P 5608--5628 %U https://proceedings.mlr.press/v235/cao24c.html %V 235 %X Multimodal fusion is crucial in joint decision-making systems for rendering holistic judgments. Since multimodal data changes in open environments, dynamic fusion has emerged and achieved remarkable progress in numerous applications. However, most existing dynamic multimodal fusion methods lack theoretical guarantees and easily fall into suboptimal problems, yielding unreliability and instability. To address this issue, we propose a Predictive Dynamic Fusion (PDF) framework for multimodal learning. We proceed to reveal the multimodal fusion from a generalization perspective and theoretically derive the predictable Collaborative Belief (Co-Belief) with Mono- and Holo-Confidence, which provably reduces the upper bound of generalization error. Accordingly, we further propose a relative calibration strategy to calibrate the predicted Co-Belief for potential uncertainty. Extensive experiments on multiple benchmarks confirm our superiority. Our code is available at https://github.com/Yinan-Xia/PDF.
APA
Cao, B., Xia, Y., Ding, Y., Zhang, C. & Hu, Q.. (2024). Predictive Dynamic Fusion. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:5608-5628 Available from https://proceedings.mlr.press/v235/cao24c.html.

Related Material