期刊
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
卷 33, 期 6, 页码 2680-2692出版社
IEEE COMPUTER SOC
DOI: 10.1109/TKDE.2019.2955078
关键词
Optimization; Indexes; Adaptation models; Manganese; Sensitivity; Correlation; Active learning; online learning; class imbalance; budgeted query; sketching learning
类别
资金
- National Natural Science Foundation of China (NSFC) [61602185, 61502177, 61876208]
- Key Project of NSFC [61836003]
- Program for Guangdong Introducing Innovative and Enterpreneurial Teams [2017ZT07X183]
- Guangdong Provincial Scientific and Technological Funds [2018B010107001, 2017B090901008, 2017A010101011, 2017B090910005]
- Pearl River S&T Nova Programof Guangzhou [201806010081]
- Tencent AI Lab RhinoBird Focused Research Program [JR201902]
- CCF-Tencent Open Research Fund [RAGR20170105]
Online Active Learning aims to manage unlabeled datastream by selectively querying labels, but faces challenges such as limited query budget and class imbalance. Previous studies use asymmetric strategies and second-order optimization to address these challenges, while the proposed novel algorithm combines asymmetric losses and queries strategies and enhances efficiency through sketching technique. Promising results demonstrate the effectiveness and efficiency of the proposed methods.
Online Active Learning (OAL) aims to manage unlabeled datastream by selectively querying the label of data. OAL is applicable to many real-world problems, such as anomaly detection in health-care and finance. In these problems, there are two key challenges: the query budget is often limited; the ratio between classes is highly imbalanced. In practice, it is quite difficult to handle imbalanced unlabeled datastream when only a limited budget of labels can be queried for training. To solve this, previous OAL studies adopt either asymmetric losses or queries (an isolated asymmetric strategy) to tackle the imbalance, and use first-order methods to optimize the cost-sensitive measure. However, the isolated strategy limits their performance in class imbalance, while first-order methods restrict their optimization performance. In this article, we propose a novel Online Adaptive Asymmetric Active learning algorithm, based on a new asymmetric strategy (merging both asymmetric losses and queries strategies), and second-order optimization. We theoretically analyze its mistake bound and cost-sensitive metric bounds. Moreover, to better balance performance and efficiency, we enhance our algorithm via a sketching technique, which significantly accelerates the computational speed with quite slight performance degradation. Promising results demonstrate the effectiveness and efficiency of the proposed methods.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据