A Survey of Deep Active Learning Ren, Pengzhen; Xiao, Yun; Chang, Xiaojun ...
ACM computing surveys,
12/2022, Volume:
54, Issue:
9
Journal Article
Peer reviewed
Open access
Active learning (AL) attempts to maximize a model’s performance gain while annotating the fewest samples possible. Deep learning (DL) is greedy for data and requires a large amount of data supply to ...optimize a massive number of parameters if the model is to learn how to extract high-quality features. In recent years, due to the rapid development of internet technology, we have entered an era of information abundance characterized by massive amounts of available data. As a result, DL has attracted significant attention from researchers and has been rapidly developed. Compared with DL, however, researchers have a relatively low interest in AL. This is mainly because before the rise of DL, traditional machine learning requires relatively few labeled samples, meaning that early AL is rarely according the value it deserves. Although DL has made breakthroughs in various fields, most of this success is due to a large number of publicly available annotated datasets. However, the acquisition of a large number of high-quality annotated datasets consumes a lot of manpower, making it unfeasible in fields that require high levels of expertise (such as speech recognition, information extraction, medical images, etc.). Therefore, AL is gradually coming to receive the attention it is due.
It is therefore natural to investigate whether AL can be used to reduce the cost of sample annotation while retaining the powerful learning capabilities of DL. As a result of such investigations, deep active learning (DeepAL) has emerged. Although research on this topic is quite abundant, there has not yet been a comprehensive survey of DeepAL-related works; accordingly, this article aims to fill this gap. We provide a formal classification method for the existing work, along with a comprehensive and systematic overview. In addition, we also analyze and summarize the development of DeepAL from an application perspective. Finally, we discuss the confusion and problems associated with DeepAL and provide some possible development directions.
Full text
Available for:
IZUM, KILJ, NUK, PILJ, SAZU, UL, UM, UPUK
Generally, data is available abundantly in unlabeled form, and its annotation requires some cost. The labeling, as well as learning cost, can be minimized by learning with the minimum labeled data ...instances. Active learning (AL), learns from a few labeled data instances with the additional facility of querying the labels of instances from an expert annotator or oracle. The active learner uses an instance selection strategy for selecting those critical query instances, which reduce the generalization error as fast as possible. This process results in a refined training dataset, which helps in minimizing the overall cost. The key to the success of AL is query strategies that select the candidate query instances and help the learner in learning a valid hypothesis. This survey reviews AL query strategies for classification, regression, and clustering under the pool-based AL scenario. The query strategies under classification are further divided into: informative-based, representative-based, informative- and representative-based, and others. Also, more advanced query strategies based on reinforcement learning and deep learning, along with query strategies under the realistic environment setting, are presented. After a rigorous mathematical analysis of AL strategies, this work presents a comparative analysis of these strategies. Finally, implementation guide, applications, and challenges of AL are discussed.
Recent successes in learning-based image classification, however, heavily rely on the large number of annotated training samples, which may require considerable human effort. In this paper, we ...propose a novel active learning (AL) framework, which is capable of building a competitive classifier with optimal feature representation via a limited amount of labeled training instances in an incremental learning manner. Our approach advances the existing AL methods in two aspects. First, we incorporate deep convolutional neural networks into AL. Through the properly designed framework, the feature representation and the classifier can be simultaneously updated with progressively annotated informative samples. Second, we present a cost-effective sample selection strategy to improve the classification performance with less manual annotations. Unlike traditional methods focusing on only the uncertain samples of low prediction confidence, we especially discover the large amount of high-confidence samples from the unlabeled set for feature learning. Specifically, these high-confidence samples are automatically selected and iteratively assigned pseudolabels. We thus call our framework cost-effective AL (CEAL) standing for the two advantages. Extensive experiments demonstrate that the proposed CEAL framework can achieve promising results on two challenging image classification data sets, i.e., face recognition on the cross-age celebrity face recognition data set database and object categorization on Caltech-256.
Deep neural network has been extensively applied to hyperspectral image (HSI) classification recently. However, its success is greatly attributed to numerous labeled samples, whose acquisition costs ...a large amount of time and money. In order to improve the classification performance while reducing the labeling cost, this article presents an active deep learning approach for HSI classification, which integrates both active learning and deep learning into a unified framework. First, we train a convolutional neural network (CNN) with a limited number of labeled pixels. Next, we actively select the most informative pixels from the candidate pool for labeling. Then, the CNN is fine-tuned with the new training set constructed by incorporating the newly labeled pixels. This step together with the previous step is iteratively conducted. Finally, Markov random field (MRF) is utilized to enforce class label smoothness to further boost the classification performance. Compared with the other state-of-the-art traditional and deep learning-based HSI classification methods, our proposed approach achieves better performance on three benchmark HSI data sets with significantly fewer labeled samples.
Outlier detection is an important topic in machine learning and has been used in a wide range of applications. In this paper, we approach outlier detection as a binary-classification issue by ...sampling potential outliers from a uniform reference distribution. However, due to the sparsity of data in high-dimensional space, a limited number of potential outliers may fail to provide sufficient information to assist the classifier in describing a boundary that can separate outliers from normal data effectively. To address this, we propose a novel Single-Objective Generative Adversarial Active Learning (SO-GAAL) method for outlier detection, which can directly generate informative potential outliers based on the mini-max game between a generator and a discriminator. Moreover, to prevent the generator from falling into the mode collapsing problem, the stop node of training should be determined when SO-GAAL is able to provide sufficient information. But without any prior information, it is extremely difficult for SO-GAAL. Therefore, we expand the network structure of SO-GAAL from a single generator to multiple generators with different objectives (MO-GAAL), which can generate a reasonable reference distribution for the whole dataset. We empirically compare the proposed approach with several state-of-the-art outlier detection methods on both synthetic and real-world datasets. The results show that MO-GAAL outperforms its competitors in the majority of cases, especially for datasets with various cluster types or high irrelevant variable ratio. The experiment codes are available at: https://github.com/leibinghe/GAAL-based-outlier-detection .
Many predictive tasks require labeled data to induce classification models. The data labeling process may have a high cost. Several strategies have been proposed to optimize the selection of the most ...relevant examples, a process referred to as active learning. However, a lack of empirical studies comparing different active learning approaches across multiple datasets makes it difficult identifying the most promising strategies, or even assessing the relative gain of active learning over the trivial random selection of instances. In this study, a comprehensive comparison of active learning strategies is presented, with various instance selection criteria, different classification algorithms and a large number of datasets. The experimental results confirm the effectiveness of active learning and provide insights about the relationship between classification algorithms and active learning strategies. Additionally, ranking curves with bands are introduced as a means to summarize in a single chart the performance of each active learning strategy for different classification algorithms and datasets.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Contribution: Practical active learning stations (PALSs)-equipped classrooms function similar to prototypical active learning classrooms (ALCs). They support student collaboration and active learning ...pedagogies but at a fraction of the cost. Background: Active learning pedagogies and active learning technology are revitalizing STEM education and their use has led to an increase in student performance and satisfaction with the learning environment in postsecondary settings. An obstacle to increasing access to ALCs is the cost of constructing such learning environments. To address this challenge, a means to retrofit an existing computer laboratory into an ALC by making use of economy hardware and open-source software was devised. Intended Outcomes: In the context of an introductory sequence of programming courses (i.e., CS1 and CS2), students in a PALS-equipped classroom would perform as well as students in a prototypical ALC. Application Design: A quasi-experimental study was employed to compare the overall student performance across learning environments. Student performance was measured by the final exam score and overall course score. Throughout the study, the PALS-equipped classroom was paired five different times in head-to-head comparisons with either a prototypical ALC or a traditional classroom. Findings: The focus of the study was the potential effects of classroom type on students' final exam score and the overall course score. A statistically significant effect was found for only one measure, which was that students in the PALS classroom in CS1 scored higher on their overall course score even when accounting for demographic differences and the pretest measure. There were no other significant effects for classroom type, either on the final exam score for either course or the overall course score in CS2.