Research map / Bandits and online learning / Online convex optimization and games
Online Maximization of Non-Decomposable Test and Population Utilities
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Studies online learning for non-decomposable classification metrics (e.g., Jaccard index, F-measure) that are general functions of the confusion matrix, with regret defined on the final empirical metric.
Problem
Classification metrics such as the Jaccard index and F-measure are non-decomposable across instances, so standard online learning with per-instance losses does not directly apply; the learner must make irrevocable sequential predictions and is judged by the final empirical metric.
Approach
Frames the task as online regret minimization on the final confusion-matrix-based metric, covering both the test-time sequence and a population utility; details beyond the abstract are not available.
Experimental setup
Metrics: regret Jaccard index F-measure
Closest papers by meaning
- A General Online Algorithm for Optimizing Complex Performance Metrics (ICML 2024 · similarity 0.90)
- Non-stationary Online Learning for Curved Losses: Improved Dynamic Regret via Mixability (ICML 2025 · similarity 0.87)
- Non-stationary Projection-Free Online Learning with Dynamic and Adaptive Regret Guarantees (AAAI 2024 · similarity 0.86)
- Sequential Probability Assignment with Contexts: Minimax Regret, Contextual Shtarkov Sums, and Contextual Normalized Maximum Likelihood (NeurIPS 2024 · similarity 0.86)
- From Average Sensitivity to Small-Loss Regret Bounds under Random-Order Model (NeurIPS 2026 · similarity 0.85)
- Efficient Non-stationary Online Learning by Wavelets with Applications to Online Distribution Shift Adaptation (ICML 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.