Neural Routed Boosting: Robust Learning against Heteroscedastic Noise

Puspak Chakraborty, Arun Rajkumar
Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, PMLR 337:961-980, 2026.

Abstract

Boosting algorithms such as AdaBoost have achieved widespread success by iteratively focusing on hard-to-classify instances. However, this aggressive re-weighting mechanism makes them susceptible to label noise and overfitting. While robust variants like Conditional Boosting handle this by estimating conditional risk, they rely on global statistical approximations. We propose Neural Routed Boosting, a novel ensemble framework that addresses heteroscedastic noise through a structural approach. Neural Routed Boosting utilizes a lightweight neural network to partition the input space into geometrically coherent regions, training specialized weak learners for each. This isolates noisy subspaces, preventing them from corrupting the decision boundaries of clean regions. We prove that the composite model of a neural router and region-specific experts constitutes a valid weak learner under the boosting framework, guaranteeing training error convergence. Experimental results on synthetic and real-world datasets demonstrate that Neural Routed Boosting outperforms traditional and robust boosting baselines, including Conditional Boosting, in high-noise environments.

Cite this Paper

BibTeX
@InProceedings{pmlr-v337-chakraborty26a, title = {Neural Routed Boosting: Robust Learning against Heteroscedastic Noise}, author = {Chakraborty, Puspak and Rajkumar, Arun}, booktitle = {Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence}, pages = {961--980}, year = {2026}, editor = {Perković, Emilija and Malinsky, Daniel}, volume = {337}, series = {Proceedings of Machine Learning Research}, month = {17--21 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v337/main/assets/chakraborty26a/chakraborty26a.pdf}, url = {https://proceedings.mlr.press/v337/chakraborty26a.html}, abstract = {Boosting algorithms such as AdaBoost have achieved widespread success by iteratively focusing on hard-to-classify instances. However, this aggressive re-weighting mechanism makes them susceptible to label noise and overfitting. While robust variants like Conditional Boosting handle this by estimating conditional risk, they rely on global statistical approximations. We propose Neural Routed Boosting, a novel ensemble framework that addresses heteroscedastic noise through a structural approach. Neural Routed Boosting utilizes a lightweight neural network to partition the input space into geometrically coherent regions, training specialized weak learners for each. This isolates noisy subspaces, preventing them from corrupting the decision boundaries of clean regions. We prove that the composite model of a neural router and region-specific experts constitutes a valid weak learner under the boosting framework, guaranteeing training error convergence. Experimental results on synthetic and real-world datasets demonstrate that Neural Routed Boosting outperforms traditional and robust boosting baselines, including Conditional Boosting, in high-noise environments.} }
Endnote
%0 Conference Paper %T Neural Routed Boosting: Robust Learning against Heteroscedastic Noise %A Puspak Chakraborty %A Arun Rajkumar %B Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2026 %E Emilija Perković %E Daniel Malinsky %F pmlr-v337-chakraborty26a %I PMLR %P 961--980 %U https://proceedings.mlr.press/v337/chakraborty26a.html %V 337 %X Boosting algorithms such as AdaBoost have achieved widespread success by iteratively focusing on hard-to-classify instances. However, this aggressive re-weighting mechanism makes them susceptible to label noise and overfitting. While robust variants like Conditional Boosting handle this by estimating conditional risk, they rely on global statistical approximations. We propose Neural Routed Boosting, a novel ensemble framework that addresses heteroscedastic noise through a structural approach. Neural Routed Boosting utilizes a lightweight neural network to partition the input space into geometrically coherent regions, training specialized weak learners for each. This isolates noisy subspaces, preventing them from corrupting the decision boundaries of clean regions. We prove that the composite model of a neural router and region-specific experts constitutes a valid weak learner under the boosting framework, guaranteeing training error convergence. Experimental results on synthetic and real-world datasets demonstrate that Neural Routed Boosting outperforms traditional and robust boosting baselines, including Conditional Boosting, in high-noise environments.
APA
Chakraborty, P. & Rajkumar, A.. (2026). Neural Routed Boosting: Robust Learning against Heteroscedastic Noise. Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 337:961-980 Available from https://proceedings.mlr.press/v337/chakraborty26a.html.

Related Material