Get Started
Home
Topics
Search
Library
Research questionHow can deep imbalanced regression avoid underfitting scarce tail labels when uncertainty varies across instances?Standard point-loss regression treats samples as having similar uncertainty, while heteroscedastic objectives can couple mean and variance learning. In long-tailed data, this can weaken optimization signals for scarce tail examples and leave them underfit.
AI
Computer Vision
Machine Learning
Research Paper
Statistical Machine Learning
Latest papersRecent research connected to this question, newest first.Beyond Homoscedasticity: Decoupled Uncertainty Optimization for Deep Imbalanced RegressionThe source studies deep imbalanced regression for tasks including age estimation, depth prediction, and protein mutation activity prediction. Its proposed framework models conditional Gaussian predictions, decouples mean and variance optimization, and uses distribution-guided contrastive learning; results are reported on IMDB-WIKI-DIR, AgeDB-DIR, and AAV2-DIR using few-shot bMAE, GM, and MAE.research paper · Sep 4, 2026
Related questions
How can probabilistic regression provide valid uncertainty estimates for unbounded outcomes?How can regression models produce tight, valid prediction intervals across confidence levels without retraining?How should imbalanced multi-label data be sampled to support rare-label inference while accounting for target frequencies and label dependencies?How should multi-expert models balance diversity and prediction variance for long-tailed visual recognition?