检索结果-内蒙古大学图书馆

A Refinement Framework of CS-MRI Using the Fast Low-Rank Approximation

学校读者我要写书评

暂无评论

A Refinement Framework of CS-MRI Using the Fast Low-Rank App...

International Conference on computer and Communications (ICCC)

作者： Shuqi Xu Youjun Xiang Xia Zeng Wei Zhang Chunsheng Zhuang Zhen Chen School of Electronic and Information Engineering Macau University of Science and Technology Macau China School of Electronic and Information Engineering South China University of Technology Guangzhou China Department of of electrical and computer engineering Guangzhou University Sontan College Guangzhou China Henan Academy of Science Applied Physics Institute Co. Ltd Henan China Institute of Microelectronics University of Macau Macau China

ISBN: (数字)9798331507077

ISBN: (纸本)9798331507084

Compressed sensing magnetic resonance imaging (CS-MRI) enables fast acquisition, which is highly desirable for numerous clinical applications. However, the quality of the MR images will depend upon the rate of undersampling. Especially, under the extremely undersampling case, if the image obtained by a conventional CS-MRI method is not satisfied, is any remedial measure could be made? In this paper, a refinement framework of CS-MRI will be proposed as the remedial measure. In the refinement framework, the MR image obtained by using the conventional CS-MRI method, will be taken as the training image. A low-rank model associated with a self-learning scheme will be established for the refinement framework. To preserve more detail features and to reduce the complexity, a partial singular value thresholding (SVT) method and an adaptive selection of the regularization parameter will be employed to establish the solver. Numerical simulations show that the proposed framework can be the remedial measure to improve the quality and robustness of MR image reconstruction in the low sampling case.

关键词： Training Thresholding (Imaging) Dictionaries Magnetic resonance imaging Image edge detection Numerical simulation Robustness Numerical models Image reconstruction Compressed sensing

Topological superconductors from a materials perspective

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Mandal, Manasi Drucker, Nathan C. Siriviboon, Phum Nguyen, Thanh Boonkird, Tiya Lamichhane, Tej Nath Okabe, Ryotaro Chotrattanapituk, Abhijatmedhi Li, Mingda Quantum Measurement Group MIT CambridgeMA02139 United States Department of Nuclear Science and Engineering MIT CambridgeMA02139 United States School of Engineering and Applied Sciences Harvard University CambridgeMA02138 United States Department of Physics MIT CambridgeMA02139 United States Department of Chemistry MIT CambridgeMA02139 United States Department of Electrical Engineering and Computer Science MIT CambridgeMA02139 United States

Topological superconductors (TSCs) have garnered significant research and industry attention in the past two decades. By hosting Majorana bound states which can be used as qubits that are robust against local perturbations, TSCs offer a promising platform toward (non-universal) topological quantum computation. However, there has been a scarcity of TSC candidates, and the experimental signatures that identify a TSC are often elusive. In this perspective, after a short review of the TSC basics and theories, we provide an overview of the TSC materials candidates, including natural compounds and synthetic material systems. We further introduce various experimental techniques to probe TSC, focusing on how a system is identified as a TSC candidate, and why a conclusive answer is often challenging to draw. We conclude by calling for new experimental signatures and stronger computational support to accelerate the search for new TSC candidates. © 2023, CC BY.

关键词： Quantum computers

On the Utilization of Unique Node Identifiers in Graph Neural Networks

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Bechler-Speicher, Maya Eliasof, Moshe Schönlieb, Carola-Bibiane Gilad-Bachrach, Ran Globerson, Amir Blavatnik School of Computer Science Tel-Aviv University Department of Applied Mathematics and Theoretical Physics University of Cambridge United Kingdom Department of Bio-Medical Engineering Tel-Aviv University Edmond J. Safra Center for Bioinformatics Tel-Aviv University Meta United States Google Research United States

Graph neural networks have inherent representational limitations due to their message-passing structure. Recent work has suggested that these limitations can be overcome by using unique node identifiers (UIDs). Here we argue that despite the advantages of UIDs, one of their disadvantages is that they lose the desirable property of permutation-equivariance. We thus propose to focus on UID models that are permutation-equivariant, and present theoretical arguments for their advantages. Motivated by this, we propose a method to regularize UID models towards permutation equivariance, via a contrastive loss. We empirically demonstrate that our approach improves generalization and extrapolation abilities while providing faster training convergence. On the recent BREC expressiveness benchmark, our proposed method achieves state-of-the-art performance compared to other random-based approaches. © 2024, CC BY.

关键词： Graph neural networks

Optimized Security Mechanism for publicly Secret Key Sharing over Cloud using Blockchain

学校读者我要写书评

暂无评论

Optimized Security Mechanism for publicly Secret Key Sharing...

2024 International Symposium on Securing Next Generation Systems Using Future Artificial Intelligence Technologies, SNSFAIT 2024

作者： Byeon, Haewon Quraishi, Aadam Thuniki, Praveen Keshta, Ismail Soni, Mukesh Shabaz, Mohammad Department of AI and Software Inje University Gimhae50834 Korea Republic of Inje University Medical Big Data Research Center Gimhae50834 Korea Republic of Intervention Treatment Institute HoustonTX United States Sr Program Analyst Georgia CummingGA30040 United States Computer Science and Information Systems Department College of Applied Sciences Al Maarefa University Riyadh Saudi Arabia Dr. D. Y. Patii Vidyapeeth Pune Dr. D. Y. Patii School of Science & Technology Tathawade Pune India Model Institute of Engineering and Technology J&K Jammu India

A publicly verifiable key sharing mechanism based on threshold key sharing is provided to explore the security of users' private keys on the blockchain. Participating nodes check the key fragment after receiving it, effectively preventing it from being abused. The crucial sections of the nodes that participated in the critical splicing are made public during the critical recovery stage to prevent them from performing harmful things during the critical recovery stage. Add IDs to the nodes that participated in the crucial splicing during the key distribution stage;a dynamic threshold system is intended to track and update the status of malicious nodes in real time. When the node that possesses the crucial component fails, the owner of the critical component and the main node relocate a key element to the new participating nodes. To safeguard sensitive information. The experimental results show that this system has a key recovery rate of 80% and threshold qualities such as traceability, enforceability, and recoverability. © 2024 Copyright for this paper by its authors.

关键词： Blockchain Encryption Key sharing mechanism Privacy Secret key Security

Inverted-circuit zero-noise extrapolation for quantum gate error mitigation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Koenig, Kathrin F. Reinecke, Finn Hahn, Walter Wellens, Thomas Fraunhofer Institute for Applied Solid State Physics Tullastr. 72 Freiburg79108 Germany Department of Sustainable Systems Engineering University of Freiburg Emmy-Noether-Str. 2 Freiburg79110 Germany Department of Computer Science University of Freiburg Georges-Köhler-Allee 51 Freiburg79110 Germany

A common approach to deal with gate errors in modern quantum-computing hardware is zero-noise extrapolation. By artificially amplifying errors and extrapolating the expectation values obtained with different error strengths towards the zero-error (zero-noise) limit, the technique aims at rectifying errors in noisy quantum computing systems. For accurate extrapolation, it is essential to know the exact factors of the noise amplification. In this article, we propose a simple method for estimating the strength of errors occurring in a quantum circuit and demonstrate improved extrapolation results. The method determines the error strength for a circuit by appending to it the inverted circuit and measuring the probability of the initial state. The estimation of error strengths is easy to implement for arbitrary circuits and does not require previous characterization of noise properties. We compare this method with the conventional zero-noise extrapolation method and show that our method leads to a more accurate calculation of expectation values on current quantum hardware, showcasing its suitability for near-term quantum computing applications. © 2024, CC BY.

关键词： Quantum computers

ENHANCING PRINTED CIRCUIT BOARD DEFECT DETECTION THROUGH ENSEMBLE LEARNING

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Law, Ka Nam Canaan Yu, Mingshuo Zhang, Lianglei Zhang, Yiyi Xu, Peng Liu, Jun Gao, Jerry Department of Applied Data Science San José State University San José United States Dept. of ISE Virginia Tech Blacksburg United States Dept. of Computer Engineering San José State University San José United States

The quality control of printed circuit boards (PCBs) is paramount in advancing electronic device technology. While numerous machine learning methodologies have been utilized to augment defect detection efficiency and accuracy, previous studies have predominantly focused on optimizing individual models for specific defect types, often overlooking the potential synergies between different approaches. This paper introduces a comprehensive inspection framework leveraging an ensemble learning strategy to address this gap. Initially, we utilize four distinct PCB defect detection models utilizing state-of-the-art methods: EfficientDet, MobileNet SSDv2, Faster RCNN, and YOLOv5. Each method is capable of identifying PCB defects independently. Subsequently, we integrate these models into an ensemble learning framework to enhance detection performance. A comparative analysis reveals that our ensemble learning framework significantly outperforms individual methods, achieving a 95% accuracy in detecting diverse PCB defects. These findings underscore the efficacy of our proposed ensemble learning framework in enhancing PCB quality control processes. © 2024, CC BY.

关键词： Adversarial machine learning

Evaluating the Effect of Volatile Federated Timeseries on Modern DNNs: Attention over Long/Short Memory

学校读者我要写书评

暂无评论

Evaluating the Effect of Volatile Federated Timeseries on Mo...

International Conference on Modern Circuits and Systems Technologies (MOCAST)

作者： Ilias Siniosoglou Konstantinos Xouveroudis Vasileios Argyriou Thomas Lagkas Sotirios K. Goudos Konstantinos E. Psannis Panagiotis Sarigiannidis Department of Electrical and Computer Engineering University of Western Macedonia Kozani Greece R&D Department MetaMind Innovations P.C. Kozani Greece Department of Networks and Digital Media Kingston University Kingston Upon Thames United Kingdom Department of Computer Science International Hellenic University Greece Physics Department Aritostle University of Thessaloniki Thessaloniki Greece Department of Applied Informatics School of Information Sciences University of Macedonia Thessaloniki Greece

In regards to the field of trend forecasting in time series, many popular Deep Learning (DL) methods such as Long Short Term Memory (LSTM) models have been the gold standard for a long time. However, depending on the domain and application, it has been shown that a new approach can be implemented and possibly be more beneficial, the Transformer deep neural networks. Moreover, one can incorporate Federated Learning (FL) in order to further enhance the prospective utility of the models, enabling multiple data providers to jointly train on a common model, while maintaining the privacy of their data. In this paper, we use an experimental Federated Learning System that employs both Transformer and LSTM models on a variety of datasets. The sytem receives data from multiple clients and uses federation to create an optimized global model. The potential of Federated Learning in real-time forecasting is explored by comparing the federated approach with conventional local training. Furthermore, a comparison is made between the performance of the Transformer and its equivalent LSTM in order to determine which one is more effective in each given domain, which shows that the Transformer model can produce better results, especially when optimised by the FL process.

关键词：

Bayesian autoencoders for data-driven discovery of coordinates, governing equations and fundamental constants

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Gao, L. Mars Kutz, J. Nathan Paul G. Allen School of Computer Science & Engineering University of Washington SeattleWA United States Department of Applied Mathematics and Electrical and Computer Engineering University of Washington SeattleWA United States

Recent progress in autoencoder-based sparse identification of nonlinear dynamics (SINDy) under `1 constraints allows joint discoveries of governing equations and latent coordinate systems from spatio-temporal data, including simulated video frames. However, it is challenging for `1-based sparse inference to perform correct identification for real data due to the noisy measurements and often limited sample sizes. To address the data-driven discovery of physics in the low-data and high-noise regimes, we propose Bayesian SINDy autoencoders, which incorporate a hierarchical Bayesian sparsifying prior: Spike-and-slab Gaussian Lasso. Bayesian SINDy autoencoder enables the joint discovery of governing equations and coordinate systems with a theoretically guaranteed uncertainty estimate. To resolve the challenging computational tractability of the Bayesian hierarchical setting, we adapt an adaptive empirical Bayesian method with Stochatic gradient Langevin dynamics (SGLD) which gives a computationally tractable way of Bayesian posterior sampling within our framework. Bayesian SINDy autoencoder achieves better physics discovery with lower data and fewer training epochs, along with valid uncertainty quantification suggested by the experimental studies. The Bayesian SINDy autoencoder can be applied to real video data, with accurate physics discovery which correctly identifies the governing equation and provides a close estimate for standard physics constants like gravity g, for example, in videos of a pendulum. Copyright © 2022, The Authors. All rights reserved.

关键词： Dynamical systems

Neuro-Tech Revolution: AI-EEG Integration for Efficient Depression Diagnosis

学校读者我要写书评

暂无评论

TechRxiv

TechRxiv 2023年

作者： Gabdrakhimov, Bekarys Detlefsen, Nicki Skafte Uyanik, Cihan Ejaz, Osama Khan, Muhammed Ahmed Hasan, Muhammad Abul Qazi, Saad Ahmed Puthusserypady, Sadasivan The Department of Health Technology Technical University of Denmark Kgs. Lyngby2800 Denmark The Department of Applied Mathematics and Computer Science Technical University of Denmark Kgs. Lyngby2800 Denmark The Department of Electrical Engineering Stanford University Palo Alto94304 United States The Neurocomputation Lab NED University of Engineering and Technology Karachi75600 Pakistan

Major depressive disorder (MDD) is a common mental disorder affecting the lives of about 280 million people and increasing rates of suicidal mortality. The current methods of diagnosis of depression are subjective, time-consuming, expensive, and inaccurate because of its heterogeneous symptoms that overlap with other disorders. In this paper, we exploit the potential of the fusion of artificial intelligence (AI) and electroencephalogram (EEG) to revolutionize the automatic diagnosis of depression and compare the classification performance of machine learning (ML) and deep learning (DL) based techniques. Results from the analysis of data recorded from 46 subjects (23 MDD and 23 Control) show that the ML methods, particularly the ensemble model with the Dempster-Shafer combination rule outperforms other models, achieving an accuracy of 99.62% and showing robustness to the variations in the data. Our work also includes a study on the effect of various hyper-parameters, in particular the number of EEG channels, feature selection methods, number of selected features, and segmentation length on the model performance. The AI-EEG integration can enhance the accuracy of diagnosis, enable personalized treatment plans, and improve patient outcomes. Continued research, development, and validation of AI algorithms, in conjunction with ethical considerations, will be crucial to harness the full potential of this technology in mental healthcare. © 2023, CC BY.

关键词： Electroencephalography