检索结果-内蒙古大学图书馆

Post training uncertainty calibration of deep networks for medical image segmentation

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Rousseau, Axel-Jan Becker, Thijs Bertels, Jeroen Blaschko, Matthew B. Valkenborg, Dirk I-Biostat Data Science Institute Hasselt University Belgium Processing Speech and Images Department of Electrical Engineering KU Leuven Belgium

Neural networks for automated image segmentation are typically trained to achieve maximum accuracy, while less attention has been given to the calibration of their confidence scores. However, well-calibrated confidence scores provide valuable information towards the user. We investigate several post hoc calibration methods that are straightforward to implement, some of which are novel. They are compared to Monte Carlo (MC) dropout. They are applied to neural networks trained with cross-entropy (CE) and soft Dice (SD) losses on BraTS 2018 and ISLES 2018. Surprisingly, models trained on SD loss are not necessarily less calibrated than those trained on CE loss. In all cases, at least one post hoc method improves the calibration. There is limited consistency across the results, so we can’t conclude on one method being superior. In all cases, post hoc calibration is competitive with MC dropout. Although average calibration improves compared to the base model, subject-level variance of the calibration remains similar. Copyright © 2020, The Authors. All rights reserved.

关键词： Calibration

Insight-HXMT observations of the extremely bright GRB 221009A

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Zhang, Wen-Long Xue, Wang-Chen Li, Cheng-Kui Xiong, Shao-Lin Li, Gang Chen, Yong Cui, Wei-Wei Li, Xiao-Bo Liu, Cong-Zhan Ge, Ming-Yu Tan, Wen-Jun Liu, Jia-Cong Wang, Chen-Wei Zheng, Chao Zhang, Yan-Qiu Wang, Yue Zhang, Zhen Yi, Shu-Xu Xiao, Shuo Cai, Ce Yi, Shuang-Xi Song, Li-Ming Tao, Lian Zhang, Shu Zhang, Shuang-Nan Key Laboratory of Particle Astrophysics Institute of High Energy Physics Chinese Academy of Sciences Beijing100049 China School of Physics and Physical Engineering Qufu Normal University Shandong Qufu273165 China University of Chinese Academy of Sciences Chinese Academy of Sciences Beijing100049 China School of Physics and Electronic Science Guizhou Normal University Guiyang550001 China Guizhou Provincial Key Laboratory of Radio Astronomy and Data Processing Guizhou Normal University Guiyang550001 China College of Physics Hebei Key Laboratory of Photophysics Research and Application Hebei Normal University Hebei Shijiazhuang050024 China

The Hard X-ray Modulation Telescope (Insight-HXMT) detected GRB 221009A, the brightest gamma-ray burst observed to date, with all its three telescopes, i.e. High Energy telescope (HE, 20-250 keV), Medium Energy telescope (ME, 5-30 keV), and Low Energy telescope (LE, 1-10 keV). Here we present the detailed observation results of all three telescopes of Insight-HXMT on the prompt emission of GRB 221009A. After dead-time and data saturation correction, we recovered the light curves of HE, ME and LE telescopes and find that they generally track the GECAM-C low gain light curves that are free of data saturation issues. Particularly, the ME light curve matches the GECAM-C light curve in low gain mode above 400 keV, while the LE light curve is more consistent with the GECAM-C above 1.5 MeV. Based on simulation, we find that the signals recorded by the ME and LE are actually caused by the secondary particles produced by the interaction between GRB gamma-ray photons and the material of the satellite. Interestingly, the consistency between ME and LE light curves and GECAM-C demonstrates that ME and LE data could be used to characterize the GRB properties. Espeically, the high time resolution light curve of ME allowed us, for the first time, to calculate the minimum variability timescale (MVT = 0.10 s) of the main burst episode of GRB 221009A. Copyright © 2025, The Authors. All rights reserved.

关键词： Gamma radiography

The Mu3e experiment: Toward the construction of an HV-MAPS vertex detector

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Rudzki, Thomas Augustin, Heiko Deflorin, Marin Dittmeier, Sebastian Frauen, Florian Immig, David M. Kim, Dohun Aeschbacher, Frank Meier González, Annie Meneses Menzel, Marius Perić, Ivan Preuß, Sebastian Schöning, André Vigani, Luigi Weber, Alena Weinläder, Benjamin Paul Scherrer Institut Villigen5232 Switzerland Universität Heidelberg Physikalisches Institut Heidelberg69120 Germany Karlsruhe Institute of Technology Institute of Data Processing and Electronics Karlsruhe76021 Germany Fachhochschule Nordwestschweiz Institut für Thermo- und Fluid-Engineering Windisch5210 Switzerland

The Mu3e experiment searches for the lepton flavor violating decay µ+ → e+ e+ e− with an ultimate aimed sensitivity of 1 event in 1016 decays. This goal can only be achieved by reducing the material budget per tracking layer to X/X0 ≈ 0.1 %. High-Voltage Monolithic Active Pixel Sensors (HV-MAPS) which are thinned to 50 µm serve as sensors. Gaseous helium is chosen as coolant. Results of recent studies related to the sensor prototypes, the helium cooling, and module prototyping are presented. The recent chip submission MuPix10 has proven its functionality regarding efficiency and time resolution. The helium cooling system for the inner tracker could be verified using a full-scale prototype. A complete prototype equipped with MuPix10 chips will be tested inside the Mu3e magnet in summer 2021. © 2021, CC BY-NC-ND.

关键词： Cooling

The Importance of Gender Specification for Detection of Driver Fatigue using a Single EEG Channel

学校读者我要写书评

暂无评论

The Importance of Gender Specification for Detection of Driv...

Biomedical engineering International Conference (BMEiCON)

作者： Mohammad Shahbakhti Matin Beiramvand Erfan Nasiri Wei Chen Jordi Solé-Casals Michal Wierzchon Anna Broniec-Wójcik Piotr Augustyniak Vaidotas Marozas Biomedical Engineering Institute Kaunas University of Technology Kaunas Lithuania Faculty of Information Technology and Communication Tampere University Tempere Finland Faculty of Statistics Mathematics and Computer Allameh Tabataba’i University Tehran Iran Center for Intelligent Medical Electronics School of Information Science and Technology the Human Phenome Institute Fudan University Shanghai China Data and Signal Processing Research Group University of Vic-Central University of Catalonia Vic Spain Institute of Psychology Jagiellonian University Krakow Poland Department of Biocybernetics and Biomedical Engineering AGH University of Science and Technology Krakow Poland

ISBN: (纸本)9781665489041

Although detection of the driver fatigue using a single electroencephalography (EEG) channel has been addressed in literature, the gender differentiation for applicability of the model has not been investigated heretofore. Motivated accordingly, we address the detection of driver fatigue based the gender-segregated datasets, where each of them contains 8 subjects. After splitting the EEG signal into its sub-bands (delta, theta, alpha, beta, and gamma) using discrete wavelet transform, the log energy entropy of each band is computed to form the feature vector. Afterwards, the feature vector is randomly split into 50% for training and 50% for the unseen testing, and fed to a support vector machine model. When comparing the classification results of fatigue driving detection between the gender segregated and non-gender segregated datasets, the former achieved the accuracy 78% and 77% for male and female subjects, respectively, than the latter (71%). The obtained results show the importance of gender-specification for the driver fatigue detection.

关键词： Training Support vector machines Computational modeling Transforms Fatigue Brain modeling Electroencephalography

Phase-Aware Complex-Cycle-Consistent Self-Supervised Learning for Monaural Speech Enhancement

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Li, Yi Sun, Yang Wang, Wenwu Naqvi, Syed Mohsen The Security Lancaster InfoLab Lancaster University LancasterLA1 4WA United Kingdom The Intelligent Sensing and Communications Group School of Engineering Newcastle University Newcastle upon TyneNE1 7RU United Kingdom The Big Data Institute University of Oxford OxfordOX3 7LF United Kingdom The Centre for Vision Speech and Signal Processing University of Surrey GuildfordGU2 7XH United Kingdom

Supervised learning for monaural speech enhancement requires a large amount of carefully labelled data to train the model. Self-supervised learning (SSL) is a technique introduced recently for mitigating this requirement. However, in these SSL-based methods, the phase information from the time-frequency representation was not used, which could lead to performance degradation in speech enhancement. In this paper, we propose a phase-aware SSL based monaural speech enhancement method. The latent representations of both the amplitude and the phase are independently studied in two decoders of the proposed speech autoencoder (SAE) with only a limited set of unseen clean speech signals. Then, the mixture autoencoder (MAE) is proposed to learn a shared latent space between the SAE and the MAE representations. To minimize the reconstruction loss between the learned amplitude and phase i.e., complex-valued information, a complex-cycle-consistent (CCC) mechanism is proposed. To further boost the enhancement performance, we also propose multi-resolution spectra losses. The proposed method is evaluated on the NOISEX and DAPS corpora, and the results show that the proposed method outperforms the state-of-the-art methods. © 2023, The Authors. All rights reserved.

关键词： Speech enhancement

Correction: A semi‑supervised interactive algorithm for change point detection

学校读者我要写书评

暂无评论

data Mining and Knowledge Discovery 2024年第2期38卷 652-652页

作者： Cao, Zhenxiang Seeuws, Nick De Vos, Maarten Bertrand, Alexander STADIUS Center for Dynamical Systems Signal Processing and Data Analytics Department of Electrical Engineering KU Leuven Leuven Belgium Department of Development and Regeneration KU Leuven Leuven Belgium Leuven.AI-KU Leuven institute for AI Leuven Belgium

Learning from pseudo lesion: A self-supervised framework for covid-19 diagnosis

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Li, Zhongliang Jin, Zhihao Li, Xuechen Shen, Linlin Computer Vision Institute College of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University Shenzhen518060 China Guangdong Key Laboratory of Intelligent Information Processing Shenzhen University Shenzhen518060 China

The Coronavirus disease 2019 (COVID-19) has rapidly spread all over the world since its first report in December 2019 and thoracic computed tomography (CT) has become one of the main tools for its diagnosis. In recent years, deep learning-based approaches have shown impressive performance in myriad image recognition tasks. However, they usually require a large number of annotated data for training. Inspired by Ground Glass Opacity (GGO), a common finding in COIVD-19 patient's CT scans, we proposed in this paper a novel self-supervised pretraining method based on pseudo lesions generation and restoration for COVID-19 diagnosis. We used Perlin noise, a gradient noise based mathematical model, to generate lesion-like patterns, which were then randomly pasted to the lung regions of normal CT images to generate pseudo COVID-19 images. The pairs of normal and pseudo COVID-19 images were then used to train an encoder-decoder architecture based U-Net for image restoration, which does not require any labelled data. The pretrained encoder was then fine-tuned using labelled data for COVID-19 diagnosis task. Two public COVID- 19 diagnosis datasets made up of CT images were employed for evaluation. Comprehensive experimental results demonstrated that the proposed self-supervised learning approach could extract better feature representation for COVID-19 diagnosis and the accuracy of the proposed method outperformed the supervised model pretrained on large scale images by 6.57% and 3.03% on SARS-CoV-2 dataset and Jinan COVID-19 dataset, respectively. © 2021, CC BY-NC-SA.

关键词： Coronavirus

Low-rank reduced biquaternion tensor ring decomposition and tensor completion

学校读者我要写书评

暂无评论

Applied Mathematics and Computation 2025年 504卷

作者： Luo, Hui Liu, Xin Liu, Wei Zhang, Yang School of Computer Science and Engineering Faculty of Innovation Engineering Macau University of Science and Technology China Macau Institute of Systems Engineering Faculty of Innovation Engineering Macau University of Science and Technology China School of Artificial Intelligence at Sun Yat-sen University the Guangdong Key Laboratory of Big Data Analysis and Processing Guangzhou China Department of Mathematics University of Manitoba MB Canada

We define the reduced biquaternion tensor ring (RBTR) decomposition and provide a detailed exposition of the corresponding algorithm RBTR-SVD. Leveraging RBTR decomposition, we propose a novel low-rank tensor completion algorithm RBTR-TV integrating RBTR ranks with total variation (TV) regularization to optimize the process. Numerical experiments on color image and video completion tasks indicate the advantages of our method. © 2025 Elsevier Inc.

关键词： Numerical methods

A novel TomoSAR imaging method with few observations based on nested array

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Jiang, Pengyu Zhang, Zhe Zhang, Bingchen Xu, Zhongqiu Key Laboratory of Technology in Geo-Spatial Information Processing and Application System Chinese Academy of Sciences Beijing100190 China Aerospace Information Research Institute Chinese Academy of Sciences Beijing100094 China School of Electronic Electrical and Communication Engineering University of Chinese Academy of Sciences Beijing100049 China Suzhou Aerospace Information Research Institute Suzhou215000 China Key Laboratory of Intelligent Aerospace Big Data Application Technology Suzhou21500 China

Synthetic aperture radar tomography (TomoSAR) baseline optimization technique is capable of reducing system complexity and improving the temporal coherence of data, which has become an important research in the field of TomoSAR. In this paper, we propose a nested TomoSAR technique, which introduces the nested array into TomoSAR as the baseline configuration. This technique obtains uniform and continuous difference co-array through nested array to increase the degrees of freedom (DoF) of the system and expands the virtual aperture along the elevation direction. In order to make full use of the difference co-array, covariance matrix of the echo needs to be obtained. Therefore, we propose a TomoSAR sparse reconstruction algorithm based on nested array, which uses adaptive covariance matrix estimation to improve the estimation performance in complex scenes. We demonstrate the effectiveness of the proposed method through simulated and real data experiments. Compared with traditional TomoSAR and coprime TomoSAR, the imaging results of our proposed method have a better anti-noise performance and retain more image information. Copyright © 2022, The Authors. All rights reserved.

关键词： Covariance matrix