arXiv：cs.AI· Kangping Hu, Stephen Mussmann·· 17 小时前
基于部分批次标签采样的批量贝叶斯主动学习（ParBaLS EPIG）
Batch Bayesian Active Learning with Partial Batch Label Sampling
AI 导读
研究者将主动学习的贝叶斯决策理论框架扩展至批量场景，为 EPIG 算法提出部分批次标签采样方法 ParBaLS。实验表明，在基于大预训练模型嵌入的贝叶斯逻辑回归上，ParBaLS EPIG 在多种数据集和不同预算下均取得更优性能，代码已开源。
正文
Abstract:Over the past couple of decades, many active learning acquisition functions have been proposed, leaving practitioners with an unclear choice of which to use. Bayesian active learning offers principled objectives with explainable intuition, including Expected Error Reduction (EER), Expected Predictive Information Gain (EPIG), and Bayesian Active Learning by Disagreements (BALD). A key challenge of such methods is the difficulty of extending to the batch setting, leading to either computational challenges (BatchBALD) or dramatic performance drops (top-$B$ selection). In this work, we extend active learning's Bayesian Decision Theory framework to derive Partial Batch Label Sampling (ParBaLS) for the EPIG algorithm. We show experimentally that ParBaLS EPIG gives superior performance for a variety of datasets with different budgets using Bayesian logistic regression on embeddings from large pre-trained models. Our code is available at this https URL.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML) |
| Cite as: | arXiv:2510.09877 [cs.LG] |
| (or arXiv:2510.09877v4 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2510.09877 arXiv-issued DOI via DataCite |
Submission history
From: Kangping Hu [view email]
[v1]
Fri, 10 Oct 2025 21:28:42 UTC (474 KB)
[v2]
Sat, 31 Jan 2026 18:11:47 UTC (551 KB)
[v3]
Fri, 8 May 2026 20:11:15 UTC (3,066 KB)
[v4]
Mon, 12 Oct 2026 20:36:44 UTC (3,469 KB)
来源：arXiv：cs.AI · arxiv.org