Rethinking deep active learning: Using unlabeled data at model training

Siméoni, Oriane; Budnik, Mateusz; Avrithis, Yannis; Gravier, Guillaume

doi:10.1109/icpr48806.2021.9412716

Cited by 36 publications

(25 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We call this the high budget regime, while the regime characterized by failure (or "cold start") is called the low budget regime. In the low budget regime, it has been shown that random selection outperforms most querying strategies [see Attenberg and Provost, 2010, Mittal et al, 2019, Zhu et al, 2020, Siméoni et al, 2021, Chandra et al, 2021. Different accounts have been offered to explain this observation: (i) failure to model uncertainty, which is more severe with a small budget of labels [Nguyen et al, 2015, Gal andGhahramani, 2016]; (ii) difficulty to accomplish diversity in high dimensional input spaces, where traditional metrics (such as the Euclidean distance) are not meaningful.…”

Section: Introductionmentioning

confidence: 99%

Active Learning on a Budget: Opposite Strategies Suit High and Low Budgets

Hacohen¹,

Dekel²,

Weinshall³

2022

Preprint

View full text Add to dashboard Cite

Investigating active learning, we focus on the relation between the number of labeled examples (budget size), and suitable corresponding querying strategies. Our theoretical analysis shows a behavior reminiscent of phase transition: typical points should best be queried in the low budget regime, while atypical (or uncertain) points are best queried when the budget is large. Combined evidence from our theoretical and empirical studies shows that a similar phenomenon occurs in simple classification models. Accordingly, we propose TypiClust -a deep active learning strategy suited for low budgets. In a comparative empirical investigation using a variety of architectures and image datasets, we report that in the low budget regime, TypiClust outperforms all other active learning strategies. Using TypiClust in a semi-supervised framework, the performance of competitive semi-supervised methods gets a significant boost, surpassing the state of the art.

show abstract

Section: Introductionmentioning

confidence: 99%

Active Learning on a Budget: Opposite Strategies Suit High and Low Budgets

Hacohen¹,

Dekel²,

Weinshall³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…To address the first problem, researchers have considered using generative networks for data augmentation [162] or assigning pseudo-labels to high-confidence samples in order to expand the labeled training set [166]. Some researchers have also used labeled and unlabeled datasets to combine supervised and semisupervised training across AL cycles [65,148]. In addition, previous heuristicbased AL [139] query strategies have proven to be ineffective when applied to DL [138]; therefore, for the one-by-one query strategy in classic AL, many researchers focus on the improvement of the batch sample query strategy [10,51,84,183], taking both the amount of information and the diversity of batch samples into account.…”

Section: The Necessity and Challenge Of Combining DL And Almentioning

confidence: 99%

A Survey of Deep Active Learning

Ren¹,

Xiao²,

Chang³

et al. 2020

Preprint

View full text Add to dashboard Cite

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 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 accorded the value it deserves. Although DL has made breakthroughs in various fields, most of this success is due to the 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 annotations, while retaining the powerful learning capabilities of DL. As a result of such investigations, deep active learning (DAL) has emerged. Although research on this topic is quite abundant, there has not yet been a comprehensive survey of DAL-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 DAL from an application perspective. Finally, we discuss the confusion and problems associated with DAL and provide some possible development directions.CCS Concepts: • Computing methodologies → Machine learning algorithms.

show abstract

“…Previous works have successfully managed to integrate unlabeled data into AL using self-supervised learning and semi-supervised learning. Siméoni et al [20] showed that initializing the target model with the features obtained from self-supervised pretraining gives AL a kickstart in performance. Contemporaneously, Mottaghi & Yeung [7] also used this technique in combination with a GAN based AL method and reported SOTA results on SVHN, CIFAR10, ImageNet, CelebA datasets.…”

Section: Related Workmentioning

confidence: 99%

On Initial Pools for Deep Active Learning

Chandra,

Desai,

Devaguptapu

et al. 2020

Preprint

View full text Add to dashboard Cite

Active Learning (AL) techniques aim to minimize the training data required to train a model for a given task. Pool-based AL techniques start with a small initial labeled pool and then iteratively pick batches of the most informative samples for labeling. Generally, the initial pool is sampled randomly and labeled to seed the AL iterations. While recent' studies have focused on evaluating the robustness of various query functions in AL, little to no attention has been given to the design of the initial labeled pool. Given the recent successes of learning representations in self-supervised/unsupervised ways, we propose to study if an intelligently sampled initial labeled pool can improve deep AL performance. We will investigate the effect of intelligently sampled initial labeled pools, including the use of selfsupervised and unsupervised strategies, on deep AL methods. We describe our experimental details, implementation details, datasets, performance metrics as well as planned ablation studies in this proposal. If intelligently sampled initial pools improve AL performance, our work could make a positive contribution to boosting AL performance with no additional annotation, developing datasets with lesser annotation cost in general, and promoting further research in the use of unsupervised learning methods for AL.

show abstract

Rethinking deep active learning: Using unlabeled data at model training

Cited by 36 publications

References 12 publications

Active Learning on a Budget: Opposite Strategies Suit High and Low Budgets

Active Learning on a Budget: Opposite Strategies Suit High and Low Budgets

A Survey of Deep Active Learning

On Initial Pools for Deep Active Learning

Contact Info

Product

Resources

About