Deeply Supervised Active Learning for Finger Bones Segmentation

Zhao, Ziyuan; Yang, Xin; Veeravalli, Bharadwaj; Zeng, Zeng

doi:10.1109/embc44109.2020.9176662

Cited by 16 publications

(15 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1 (a). Different from earlier work [28], we adopt a light variant of U-Net [5], in which, upsampling layers only include nearest-neighbor interpolation instead of interpolation with 1 × 1 convolution. Upon the basic architecture of U-Net [5], we inject deep supervision into several hidden layers, namely lower layer and middle layer.…”

Section: A Deeply Supervised U-netmentioning

confidence: 99%

See 1 more Smart Citation

DSAL: Deeply Supervised Active Learning From Strong and Weak Labelers for Biomedical Image Segmentation

Zhao

Zeng

et al. 2021

IEEE J. Biomed. Health Inform.

Self Cite

View full text Add to dashboard Cite

Image segmentation is one of the most essential biomedical image processing problems for different imaging modalities, including microscopy and X-ray in the Internet-of-Medical-Things (IoMT) domain. However, annotating biomedical images is knowledge-driven, timeconsuming, and labor-intensive, making it difficult to obtain abundant labels with limited costs. Active learning strategies come into ease the burden of human annotation, which queries only a subset of training data for annotation. Despite receiving attention, most of active learning methods still require huge computational costs and utilize unlabeled data inefficiently. They also tend to ignore the intermediate knowledge within networks. In this work, we propose a deep active semi-supervised learning framework, DSAL, combining active learning and semi-supervised learning strategies. In DSAL, a new criterion based on deep supervision mechanism is proposed to select informative samples with high uncertainties and low uncertainties for strong labelers and weak labelers respectively. The internal criterion leverages the disagreement of intermediate features within the deep learning network for active sample selection, which subsequently reduces the computational costs. We use the proposed criteria to select samples for strong and weak labelers to produce oracle labels and pseudo labels simultaneously at each active learning iteration in an ensemble learning manner, which can be examined with IoMT Platform. Extensive experiments on multiple medical image datasets demonstrate the superiority of the proposed method over state-of-the-art active learning methods.

show abstract

Section: A Deeply Supervised U-netmentioning

confidence: 99%

“…• RSNA Bone Age dataset [7]: includes 12611 hand radiographs. We follow the preprocessing and sampling methods from [28] and obtained a small balanced dataset where |L 0 | = 10 and |U 0 | = 129. The evaluation is carried on a striped test set of 50 samples.…”

Section: A Datasetsmentioning

confidence: 99%

DSAL: Deeply Supervised Active Learning From Strong and Weak Labelers for Biomedical Image Segmentation

Zhao

Zeng

et al. 2021

IEEE J. Biomed. Health Inform.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Tasks of the Social Media Mining for Health Applications (SMM4H) 2021 overall tasks. [20] requires participants to develop an automated classification system to identify mentions of the self-report, Non-personal reports, as well Bibliography. COVID-19 SARS news, articles, and related topics are mentioned. …”

Section: Task and Data Descriptionmentioning

confidence: 99%

Sentimental Tweets Classification of Symptomatic COVID-19

2021

Preprint

View full text Add to dashboard Cite

The approach I described is straightforward, related to COVID-19 SARS based tweets and the symptoms, that people tweet about. Also, social media mining for health application reports was shared in many different tasks of 2021. The motto at the back of this observe is to analyses tweets of COVID-19 based symptoms. By performing BERT model and text classification with XLNET with which uses to classify text and purpose of the texts (i.e.) tweets. So that I can get a deep understanding of the texts. When developing the system, I used two models the XLNet and DistilBERT for the text sorting task, but the outcome was XLNET out-performs the given approach to the best accuracy achieved. Now I discover a whole lot vital for as it should be categorizing tweets as encompassing self-said COVID-19 indications. Whether or not a tweets associated with COVID-19 is a non-public report or an information point out to the virus. Which gives test accuracy to an F1 score of 96%.

show abstract

“…7. Taking a common CNN as an example, this figure presents a comparison between the traditional uncertainty measurement method [35,103,176] and the uncertainty measurement method of synthesizing information in two stages [59,178,182] (i.e., the feature extraction stage and task learning stage).…”

Section: Common Framework Dalmentioning

confidence: 99%

“…1c illustrates a very common general framework for DAL tasks. Related works include [35,59,103,176,182] , among others. More specifically, [176] proposes a framework that uses a fully convolutional network (FCN) [101] and AL to solve the medical image segmentation problem using a small number of annotations.…”

Section: Common Framework Dalmentioning

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

Deeply Supervised Active Learning for Finger Bones Segmentation

Cited by 16 publications

References 47 publications

DSAL: Deeply Supervised Active Learning From Strong and Weak Labelers for Biomedical Image Segmentation

DSAL: Deeply Supervised Active Learning From Strong and Weak Labelers for Biomedical Image Segmentation

Sentimental Tweets Classification of Symptomatic COVID-19

A Survey of Deep Active Learning

Contact Info

Product

Resources

About