[论文解读] A Survey of Deep Active Learning

Pengzhen Ren, Xiao Yun+6 more|arXiv (Cornell University)|Aug 30, 2020

Machine Learning and Algorithms被引用 39

一句话总结

本综述对 DeepAL 进行了全面的分类与概述，详述了挑战、查询策略、架构、应用以及未来方向。

ABSTRACT

Active learning (AL) attempts to maximize the performance gain of the model by marking the fewest samples. Deep learning (DL) is greedy for data and requires a large amount of data supply to optimize massive parameters, so that the model learns how to extract high-quality features. In recent years, due to the rapid development of internet technology, we are in an era of information torrents and we have massive amounts of data. In this way, DL has aroused strong interest of researchers and has been rapidly developed. Compared with DL, researchers have relatively low interest in AL. This is mainly because before the rise of DL, traditional machine learning requires relatively few labeled samples. Therefore, early AL is difficult to reflect the value it deserves. Although DL has made breakthroughs in various fields, most of this success is due to the publicity of the large number of existing annotation datasets. However, the acquisition of a large number of high-quality annotated datasets consumes a lot of manpower, which is not allowed in some fields that require high expertise, especially in the fields of speech recognition, information extraction, medical images, etc. Therefore, AL has gradually received due attention. A natural idea is whether AL can be used to reduce the cost of sample annotations, while retaining the powerful learning capabilities of DL. Therefore, deep active learning (DAL) has emerged. Although the related research has been quite abundant, it lacks a comprehensive survey of DAL. This article is to fill this gap, we provide a formal classification method for the existing work, and a comprehensive and systematic overview. In addition, we also analyzed and summarized the development of DAL from the perspective of application. Finally, we discussed the confusion and problems in DAL, and gave some possible development directions for DAL.

研究动机与目标

• 解释为何将深度学习( DL )与主动学习( AL )结合对在高维数据中降低标注成本具有价值。
• 为 DeepAL 方法和策略提供正式的框架与分类体系。
• 总结跨领域的现有 DeepAL 应用，以识别共同模式与差距。
• 突出将 DL 与 AL 整合时的挑战并提出未来研究方向。

提出的方法

• 定义基于池的 DeepAL 框架，并把在达到目标准确率的同时减少标记数据的优化目标形式化。
• 将 DeepAL 查询策略分为 Batch Mode DeepAL (BMDAL)、基于不确定性和混合策略、深度贝叶斯主动学习（DBAL）、基于密度的方法，以及自动化 DeepAL 设计。
• 讨论如何通过贝叶斯方法、数据增强、伪标签和半监督训练来解决 DL 特定挑战，如模型不确定性、标注样本数据稀缺以及管道不一致性。
• 描述诸如 BALD 与 BatchBALD 的批量获取策略，并对比仅不确定性与关注多样性的方法（例如 Exploration-P、DMBAL、BADGE、WAAL、TA-VAAL）。
• 概述两个核心的 DeepAL 组成部分：(i) 未标记数据上的 AL 查询策略，以及 (ii) DL 模型的训练与优化过程。

(a) Structure diagram of convolutional neural network.

实验结果

研究问题

• RQ1 将传统 AL 策略应用于 DL 模型时的关键挑战是什么？
• RQ2 如何设计基于批次的查询策略，以在 DeepAL 中平衡信息含量与样本多样性？
• RQ3 哪些方法（贝叶斯、半监督、数据增强）有助于基于 DL 的 AL 应对有限的标注数据和高不确定性？
• RQ4 如何使 DeepAL 的架构和查询策略实现自动化或任务感知，以在不同领域提高性能？
• RQ5 DeepAL 的主要应用领域和未来方向是什么？

主要发现

• 基于批量的查询（BatchBALD）通过考虑样本间相关性，优于逐一 BALD。
• 基于不确定性的策略对于小批量简单且有效，但可能损害多样性和分布覆盖。
• 平衡不确定性和多样性的混合策略（如 BADGE、WAAL、TA-VAAL）在实用性上展现出显著潜力且不需要大量超参数调优。
• 贝叶斯深度学习和数据增强有助于缓解 DL 基于 AL 的过度自信和数据稀缺。
• 需要面向 DL、适用于高维数据并与跨应用的最终任务性能对齐的批量查询设计。

(b) The pool-based active learning cycle.