Greedy Confidence Pursuit: A Pragmatic Approach to Multi-bandit Optimization

Bachman, Philip; Precup, Doina

doi:10.1007/978-3-642-40988-2_16

Cited by 7 publications

(6 citation statements)

References 8 publications

(18 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Consistency-based regularization (Berthelot et al 2019b;Xie et al 2020;Laine and Aila 2017;Berthelot et al 2019a;Tarvainen and Valpola 2017) applies a consistency loss by enforcing invariance on unlabeled data under different augmentations. Pseudo-labeling relies on the model's high confident predictions to produce pseudo-labels (Lee et al 2013;Bachman, Alsharif, and Precup 2014;Arazo et al 2020) for unlabeled data and trains them jointly with labeled data. FixMatch (Sohn et al 2020a) is a combination of both consistency-based regularization and pseudolabeling approaches.…”

Section: Related Workmentioning

confidence: 99%

Semi-supervised Class-Agnostic Motion Prediction with Pseudo Label Regeneration and BEVMix

Wang,

Wu,

Pan

et al. 2024

AAAI

View full text Add to dashboard Cite

Class-agnostic motion prediction methods aim to comprehend motion within open-world scenarios, holding significance for autonomous driving systems. However, training a high-performance model in a fully-supervised manner always requires substantial amounts of manually annotated data, which can be both expensive and time-consuming to obtain. To address this challenge, our study explores the potential of semi-supervised learning (SSL) for class-agnostic motion prediction. Our SSL framework adopts a consistency-based self-training paradigm, enabling the model to learn from unlabeled data by generating pseudo labels through test-time inference. To improve the quality of pseudo labels, we propose a novel motion selection and re-generation module. This module effectively selects reliable pseudo labels and re-generates unreliable ones. Furthermore, we propose two data augmentation strategies: temporal sampling and BEVMix. These strategies facilitate consistency regularization in SSL. Experiments conducted on nuScenes demonstrate that our SSL method can surpass the self-supervised approach by a large margin by utilizing only a tiny fraction of labeled data. Furthermore, our method exhibits comparable performance to weakly and some fully supervised methods. These results highlight the ability of our method to strike a favorable balance between annotation costs and performance. Code will be available at https://github.com/kwwcv/SSMP.

show abstract

Section: Related Workmentioning

confidence: 99%

Semi-supervised Class-Agnostic Motion Prediction with Pseudo Label Regeneration and BEVMix

Wang,

Wu,

Pan

et al. 2024

AAAI

View full text Add to dashboard Cite

show abstract

“…To alleviate this problem, noise can be added to the model during the inference time to generate more accurate predictions. This method is used in the Pseudo-Ensemble Agreement [48] and has demonstrated excellent performance. Thus, a teacher model injected with noise can be inferred to generate more precise targets than that not injected with noise.…”

Section: Data Augmentationmentioning

confidence: 99%

Semi-Supervised Learning-Enhanced Fingerprint Indoor Positioning by Exploiting an Adapted Mean Teacher Model

Chen,

Liu,

et al. 2024

Electronics

View full text Add to dashboard Cite

Location awareness is crucial for numerous emerging wireless indoor applications. Deep learning algorithms have demonstrated the potential for achieving the required level of positioning accuracy in indoor environments. However, obtaining abundant labels for data-driven machine learning is costly in practical situations. As an effective solution to alleviating the insufficiency of labeled data for deep learning-based indoor positioning, deep semi-supervised learning (DSSL) can be employed to lessen the dependency on labeled data by exploiting potential patterns in unlabeled samples. In this paper, we propose an Adapted Mean Teacher (AMT) model within the DSSL paradigm for indoor fingerprint positioning by using a channel impulse response. To enhance the generalization of the trained model, we design an efficient implicit augmentation scheme for the training process in the AMT model. Furthermore, we develop a tailored residual network to efficiently extract location characteristics in the AMT framework. We conduct extensive simulation experiments for indoor scenarios with heavy non-line-of-sight conditions based on open datasets to demonstrate the effectiveness of our proposed AMT model. Numerical results indicate that the AMT model outperforms several consistency regularization methods and the pseudo-label method in terms of positioning accuracy and lower positioning latency, achieving a mean error of 90cm when using a small number of labels.

show abstract

“…Consistency based SSL has been extensively studied in the context of deep learning in recent years [10,[14][15][16]25] . These methods leverage unlabeled data by adding an unsupervised regularization term to the standard supervised loss:…”

Section: Related Workmentioning

confidence: 99%

Meta-Semi: A Meta-Learning Approach for Semi-Supervised Learning

Wang

Guo

Song³

et al. 2022

CAAI Artificial Intelligence Research

View full text Add to dashboard Cite

Radio telescopes produce visibility data about celestial objects, but these data are sparse and noisy. As a result, images created on raw visibility data are of low quality. Recent studies have used deep learning models to reconstruct visibility data to get cleaner images. However, these methods rely on a substantial amount of labeled training data, which requires significant labeling effort from radio astronomers. Addressing this challenge, we propose VisRec, a model-agnostic semi-supervised learning approach to the reconstruction of visibility data. Specifically, VisRec consists of both a supervised learning module and an unsupervised learning module. In the supervised learning module, we introduce a set of data augmentation functions to produce diverse training examples. In comparison, the unsupervised learning module in VisRec augments unlabeled data and uses reconstructions from non-augmented visibility data as pseudo-labels for training. This hybrid approach allows VisRec to effectively leverage both labeled and unlabeled data. This way, VisRec performs well even when labeled data is scarce. Our evaluation results show that VisRec outperforms all baseline methods in reconstruction quality, robustness against common observation perturbation, and generalizability to different telescope configurations.

show abstract

Greedy Confidence Pursuit: A Pragmatic Approach to Multi-bandit Optimization

Cited by 7 publications

References 8 publications

Semi-supervised Class-Agnostic Motion Prediction with Pseudo Label Regeneration and BEVMix

Semi-supervised Class-Agnostic Motion Prediction with Pseudo Label Regeneration and BEVMix

Semi-Supervised Learning-Enhanced Fingerprint Indoor Positioning by Exploiting an Adapted Mean Teacher Model

Meta-Semi: A Meta-Learning Approach for Semi-Supervised Learning

Contact Info

Product

Resources

About