aiqbench analysis
This research introduces DUO, a novel framework addressing gradient coupling issues in deep imbalanced regression by decoupling mean and variance optimization to better handle instance-wise heteroscedasticity. By modeling targets as conditional Gaussian distributions and employing distribution-guided contrastive learning, the method significantly improves performance on tail samples across visual and biological benchmarks. The approach offers a critical advancement for continuous prediction tasks where label-scarce data carries high practical value, moving beyond standard deterministic point mappings.
Beyond Homoscedasticity: Decoupled Uncertainty Optimization for Deep Imbalanced Regression
arXiv:2609.04995v1 Announce Type: new Abstract: Deep Imbalanced Regression (DIR) is pervasive in continuous prediction tasks across diverse modalities, such as age estimation, depth prediction, and protein mutation activity prediction, where label-scarce tail samples often carry higher practical value. However, most existing methods still learn deterministic point mappings under mean squared error or its simple variants, implicitly assuming a uniform uncertainty level across all samples and thereby overlooking the instance-wise heteroscedasticity that is widespread in long-tailed data. We further point out that even heteroscedastic negative log-likelihood suffers from a gradient coupling issue, which, under DIR scenarios, weakens the learning signal of hard tail samples and leads to optimization inertia as well as tail underfitting. To address this, we propose DUO, an uncertainty-aware long-tailed regression framework. Specifically, the proposed method models the regression target as a conditional Gaussian distribution to explicitly characterize instance-level predictive uncertainty, and transforms uncertainty into a dynamic enhancement signal for tail samples through decoupled mean-variance optimization. Furthermore, we design a distribution-guided contrastive learning mechanism that adaptively constructs positive and negative pairs based on the overlap between sample distributions, thereby alleviating feature looseness and cross-label semantic entanglement. Across visual and biological DIR benchmarks, DUO achieves the best few-shot bMAE and GM on IMDB-WIKI-DIR, AgeDB-DIR, and AAV2-DIR while remaining competitive on few-shot MAE.
Abstract:Deep Imbalanced Regression (DIR) is pervasive in continuous prediction tasks across diverse modalities, such as age estimation, depth prediction, and protein mutation activity prediction, where label-scarce tail samples often carry higher practical value. However, most existing methods still learn deterministic point mappings under mean squared error or its simple variants, implicitly assuming a uniform uncertainty level across all samples and thereby overlooking the instance-wise heteroscedasticity that is widespread in long-tailed data. We further point out that even heteroscedastic negative log-likelihood suffers from a gradient coupling issue, which, under DIR scenarios, weakens the learning signal of hard tail samples and leads to optimization inertia as well as tail underfitting. To address this, we propose DUO, an uncertainty-aware long-tailed regression framework. Specifically, the proposed method models the regression target as a conditional Gaussian distribution to explicitly characterize instance-level predictive uncertainty, and transforms uncertainty into a dynamic enhancement signal for tail samples through decoupled mean-variance optimization. Furthermore, we design a distribution-guided contrastive learning mechanism that adaptively constructs positive and negative pairs based on the overlap between sample distributions, thereby alleviating feature looseness and cross-label semantic entanglement. Across visual and biological DIR benchmarks, DUO achieves the best few-shot bMAE and GM on IMDB-WIKI-DIR, AgeDB-DIR, and AAV2-DIR while remaining competitive on few-shot MAE.
Accepted at the 34th ACM International Conference on Multimedia (ACM MM 2026). Main paper with supplementary material
Machine Learning (cs.LG)
arXiv:2609.04995 [cs.LG]
(or arXiv:2609.04995v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.04995 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 319 words
Related coverage
Why do models task game?
AI Alignment Forum · Aug 6, 2026
Dynamic Loss Balancing for Joint SOH and RUL Prediction of Lithium-Ion Batteries via a Rotary SOH-Injected Prior Battery Transformer
arXiv · Jul 23, 2026
Physics-Guided Masked Multi-Task Network for Edge-Friendly Battery Health Diagnostics from Sto-chastically Fragmented Charging Profiles
arXiv · Jul 23, 2026
Latent reasoning architectures would likely undermine CoT, our strongest oversight tool
AI Alignment Forum · Sep 23, 2026