Improving Dataset Volumes and Model Accuracy With Semi-Supervised Iterative Self-Learning

Dupre, Rob; Fajtl, Jiri; Argyriou, Vasileios; Remagnino, Paolo

doi:10.1109/tip.2019.2913986

Cited by 20 publications

(8 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Carmon et al (2019) proved that using unlabelled data can help to achieve high robust accuracy as well as high standard accuracy at the same time. Dupre et al (2019) considered iteratively pseudo-labelling the whole unlabelled dataset with a confidence threshold and showed that the accuracy converges relatively quickly. Oymak & Gulcu (2021), in which part of our analysis hinges on, studied SSL under the binary Gaussian mixture model setup and characterized the correlation between the learned and the optimal estimators concerning the margin and the regularization factor.…”

Section: Resultsmentioning

confidence: 99%

Information-Theoretic Characterization of the Generalization Error for Iterative Semi-Supervised Learning

He¹,

Yan²,

Tan³

2021

Preprint

View full text Add to dashboard Cite

We consider iterative semi-supervised learning (SSL) algorithms that iteratively generate pseudolabels for a large amount unlabelled data to progressively refine the model parameters. In particular, we seek to understand the behaviour of the generalization error of iterative SSL algorithms using information-theoretic principles. To obtain bounds that are amenable to numerical evaluation, we first work with a simple model-namely, the binary Gaussian mixture model. Our theoretical results suggest that when the class conditional variances are not too large, the upper bound on the generalization error decreases monotonically with the number of iterations, but quickly saturates. The theoretical results on the simple model are corroborated by extensive experiments on several benchmark datasets such as the MNIST and CIFAR datasets in which we notice that the generalization error improves after several pseudo-labelling iterations, but saturates afterwards.

show abstract

Section: Resultsmentioning

confidence: 99%

Information-Theoretic Characterization of the Generalization Error for Iterative Semi-Supervised Learning

He¹,

Yan²,

Tan³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Particularly, there have been a few works [173,75,243] involving semi-supervised learning for skin disease diagnosis. Recently, semi-supervised deep learning attracts increasing attention in the field of computer vision and a few successful models have been proposed [244,245,246]. Understanding these models and developing semi-supervised deep learning models specifically for skin disease diagnosis can be a promising direction.…”

Section: Develop Semi-supervised Deep Learning Methods For Skin Disea...mentioning

confidence: 99%

Skin disease diagnosis with deep learning: a review

Li¹,

Pan²,

Zhao³

et al. 2020

Preprint

View full text Add to dashboard Cite

Skin cancer is one of the most threatening diseases worldwide. However, diagnosing skin cancer correctly is challenging. Recently, deep learning algorithms have emerged to achieve excellent performance on various tasks. Particularly, they have been applied to the skin disease diagnosis tasks. In this paper, we present a review on deep learning methods and their applications in skin disease diagnosis. We first present a brief introduction to skin diseases and image acquisition methods in dermatology, and list several publicly available skin datasets for training and testing algorithms. Then, we introduce the conception of deep learning and review popular deep learning architectures. Thereafter, popular deep learning frameworks facilitating the implementation of deep learning algorithms and performance evaluation metrics are presented. As an important part of this article, we then review the literature involving deep learning methods for skin disease diagnosis from several aspects according to the specific tasks. Additionally, we discuss the challenges faced in the area and suggest possible future research directions. The major purpose of this article is to provide a conceptual and systematically review of the recent works on skin disease diagnosis with deep learning. Given the popularity of deep learning, there remains great challenges in the area, as well as opportunities that we can explore in the future.

show abstract

“…In the first place, there is a lack of distinction between the normal skin and the skin lesion. To make segmentation more difficult, variants in skin tones, the presence of ancient artifacts such as hair or ink on the image, air bubbles or ruler marks, non-uniform lighting conditions, the physical location of lesions and variants in lesion colour, crispness, shape, size, and position in the image all contribute to its difficulty [27]. Deep learning algorithms have been used to segment skin lesions in Table…”

Section: Deep Learning Applications In the Detection Of Skin Diseases...mentioning

confidence: 99%

Diagnosis of skin diseases based on deep learning and machine learning approach

Poornima

Sakkari

2022

ijhs

View full text Add to dashboard Cite

Skin Diseases have become one of the most frequent diseases in the world; apart from their physical effects, they are rapidly expanding to cover a bigger region and have a psychological impact on patients. Diagnosing skin diseases takes a high degree of knowledge and is objective to the dermatologist. Presented an overview of Deep-Learning algorithms and their usage in skin disease diagnostics. Begin with a short overview of skin diseases and image collecting techniques in dermatology, followed by a list of publicly accessible Skin-Datasets for algorithm training and testing are discussed. Next examine the research including deep learning algorithms for skin disease detection from many perspectives according to the particular objectives as an essential element of this paper. The main objective of this paper is to present a comprehensive and simplify analysis of contemporary deep learning-based skin disease diagnostic research. By considering the growing popularity of deep learning, there are still several challenges to overcome as well as chances to explore in the future. This study found that the diagnostic accuracy of image processing techniques was inconsistent, but the accuracy of deep Learning and Machine Learning algorithms was high, with a 99.04 % accuracy rate.

show abstract

Improving Dataset Volumes and Model Accuracy With Semi-Supervised Iterative Self-Learning

Cited by 20 publications

References 18 publications

Information-Theoretic Characterization of the Generalization Error for Iterative Semi-Supervised Learning

Information-Theoretic Characterization of the Generalization Error for Iterative Semi-Supervised Learning

Skin disease diagnosis with deep learning: a review

Diagnosis of skin diseases based on deep learning and machine learning approach

Contact Info

Product

Resources

About