One Knob to Rule Them All: A Unified Optimal Transport View of Cold-Start Active Learning

2026-08-04

AI Summary

This work addresses the challenges in cold-start active learning, where the absence of prior knowledge and the diversity of existing selection strategies hinder a unified approach. To this end, we propose the first optimal transport–based unified selection framework, which reveals the common structural underpinnings of prevailing strategies. By adaptively tuning the entropy regularization strength, our method derives a data-driven ε-AS selection rule and establishes a task-agnostic minimax generalization error bound. Empirical evaluations demonstrate that the proposed approach achieves state-of-the-art performance across six public benchmarks, yielding a 1.29% average accuracy improvement over ActiveFT on ImageNet-1k while reducing selection time by 56.2%.

Abstract

Cold-Start Active Learning (CSAL) aims to select a valuable subset from an unlabeled pool without any prior knowledge or human assistance. Existing methods take diverse routes based on typicality, coverage, or diversity. Each rests on its own inductive bias and therefore performs well on some tasks yet poorly on others. We argue that the real challenge is not to design yet another selection heuristic, but to make CSAL adapt automatically to the data and task at hand. To this end, we revisit CSAL through the lens of optimal transport. First, we propose a generalized transport selection framework that reveals the shared allocation structure of existing methods and exactly subsumes representative formulations. Second, we introduce a theoretical analysis that characterizes the trade-off controlled by entropic regularization and establishes a task-agnostic minimax bound for cold-start selection. These results provide a principled foundation for adapting the regularization strength to the unlabeled data. Third, we derive a data-adaptive regularization rule and present a novel Sinkhorn-based CSAL algorithm, termed $ε$-Adaptive Selection ($ε$-AS). Extensive experiments on six public datasets and multiple annotation budgets show that $ε$-AS consistently achieves state-of-the-art performance. On ImageNet-1k, it improves the average accuracy over ActiveFT by 1.29% while reducing selection time by 56.2%. Code will be released at https://github.com/Z-yiwei/OT-CSAL

Problem

• Cold-Start Active Learning
• Optimal Transport
• Data Adaptation
• Inductive Bias
• Unlabeled Data Selection

Innovation

• Optimal Transport
• Cold-Start Active Learning
• Entropic Regularization
• Data-Adaptive Selection
• Sinkhorn Algorithm