arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2401.09787cs.LGcs.AIstat.ML
收藏

为深度主动学习查询易翻转样本

Querying Easily Flip-flopped Samples for Deep Active Learning

KAIST(韩国科学技术院);Korea Institute of Oriental Medicine(韩国东洋医学研究院);Jeonbuk National University(全北国立大学);Kim Jaechul Graduate School of AI, KAIST(韩国科学技术院金在哲人工智能研究生院)

Seong Jin Cho, Gwangsu Kim, Junghyun Lee, Jinwoo Shin, Chang D. Yoo

AI总结:

针对多分类任务中样本到决策边界距离难以计算的问题,提出最小不一致度量LDM及其渐近一致估计器,通过查询LDM最小的未标记样本实现主动学习,在多数据集和深度架构上取得SOTA性能。

摘要:

主动学习是一种机器学习范式,旨在通过策略性地选择并查询未标记数据来提升模型性能。一种有效的选择策略基于模型的预测不确定性,该指标可被解读为样本信息量的度量。样本到决策边界的距离是预测不确定性的天然度量方式,但该值通常难以计算,尤其是在多分类任务形成的复杂决策边界场景下。为解决这一问题,本文提出了最小不一致度量(least disagree metric,LDM),其定义为预测标签的最小不一致概率,同时提出了一种LDM估计器,经证明在温和假设下具有渐近一致性。该估计器计算效率高,可通过参数扰动轻松应用于深度学习模型。基于LDM的主动学习通过查询LDM最小的未标记数据来实现。实验结果表明,本文提出的基于LDM的主动学习算法在所有测试数据集和深度架构上均取得了当前最优的整体性能。

英文摘要:

Active learning is a machine learning paradigm that aims to improve the performance of a model by strategically selecting and querying unlabeled data. One effective selection strategy is to base it on the model's predictive uncertainty, which can be interpreted as a measure of how informative a sample is. The sample's distance to the decision boundary is a natural measure of predictive uncertainty, but it is often intractable to compute, especially for complex decision boundaries formed in multiclass classification tasks. To address this issue, this paper proposes the {\it least disagree metric} (LDM), defined as the smallest probability of disagreement of the predicted label, and an estimator for LDM proven to be asymptotically consistent under mild assumptions. The estimator is computationally efficient and can be easily implemented for deep learning models using parameter perturbation. The LDM-based active learning is performed by querying unlabeled data with the smallest LDM. Experimental results show that our LDM-based active learning algorithm obtains state-of-the-art overall performance on all considered datasets and deep architectures.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑