检索结果-内蒙古大学图书馆

Transclaw U-net: Claw U-net with transformers for medical image segmentation

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Yao, Chang Hu, Menghan Zhai, Guangtao Zhang, Xiao-Ping Shanghai Key Laboratory of Multidimensional Information Processing East China Normal University Institute of Image Communication and Information Processing Shanghai Jiao Tong University Department of Electrical Computer and Biomedical Engineering Ryerson University

In recent years, computer-aided diagnosis has become an increasingly popular topic. Methods based on convolutional neural networks have achieved good performance in medical image segmentation and classification. Due to the limitations of the convolution operation, the long-term spatial features are often not accurately obtained. Hence, we propose a TransClaw U-Net network structure, which combines the convolution operation with the transformer operation in the encoding part. The convolution part is applied for extracting the shallow spatial features to facilitate the recovery of the image resolution after upsampling. The transformer part is used to encode the patches, and the self-attention mechanism is used to obtain global information between sequences. The decoding part retains the bottom upsampling structure for better detail segmentation performance. The experimental results on Synapse Multi-organ Segmentation Datasets show that the performance of TransClaw U-Net is better than other network structures. The ablation experiments also prove the generalization performance of TransClaw U-Net. © 2021, CC BY.

关键词： Convolution

Radiomics Model Based on Computed Tomography images for Prediction of Radiation-Induced Optic Neuropathy Following Radiotherapy of Brain and Head and Neck Tumors

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Nafchi, Elham Raiesi Fadavi, Pedram Amiri, Sepideh Cheraghi, Susan Garousi, Maryam Nabavi, Mansoureh Daneshi, Iman Gomar, Marzieh Molaie, Malihe Noraeinejad, Ali Department of Radiation Sciences Faculty of Allied Medicine Iran University of Medical Sciences Tehran Iran Department of Radiation Oncology School of Medicine Iran University of Medical Science Tehran Iran Department of Computer Science University of Copenhagen Copenhagen Denmark Radiation Biology Research Center Iran University of Medical Sciences Tehran Iran Cancer Institute Tehran University of Medical Sciences Tehran Iran Department of Clinical Oncology Haft-e-Tir Hospital Iran University of Medical Science Tehran Iran Control and Intelligent Processing Center of Excellence School of Electrical and Computer Engineering College of Engineering University of Tehran Tehran Iran Department of Optometry and Vision Science School of Rehabilitation Tehran University of Medical Science Tehran Iran

Purpose: We aimed to build a machine learning-based model to predict radiation-induced optic neuropathy in patients who had treated head and neck cancers with radiotherapy. Materials and Methods: To measure radiation-induced optic neuropathy, the visual evoked potential values were obtained in both case and control groups and compared. The radiomics features were extracted from the area segmented by Swin UNETR version 2, which included the right and left optic nerves and chiasm. We ranked 5 supervised ML models Bernoulli Naive Bayes, Decision Tree, Gradient Boosting Decision Trees, K-Nearest Neighbor, and Random Forest on 4 input datasets to predict radiation-induced visual complications classifiers by implementing 5-fold cross-validation. The F1 score, accuracy, sensitivity, specificity, and area under the ROC curve were compared to access prediction capability. Results: radiation-induced optic neuropathy affected 31% of the patients. 856 radiomic characteristics were extracted and selected from each segmented area. Decision Tree and Random Forest with the highest AUC (97% and 95% respectively) among the five classifiers were the most powerful algorithms to predict radiation-induced optic neuropathy. Chiasm with higher sensitivity and precision can predict radiation-induced optic neuropathy better than right or left optic nerve or combination of all radiomic features. Conclusion: We found that combination of radiomic, dosimetric, and clinical factors can predict radiation-induced optic neuropathy after radiation treatment with high accuracy. To acquire more reliable results, it is recommended to conduct visual evoked potential tests before and after radiation therapy, with multiple follow-up courses and more patients. © 2024, The Authors. All rights reserved.

关键词： Decision trees

How can spherical CNNs benefit ML-based diffusion MRI parameter estimation?

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Goodwin-Allcock, Tobias McEwen, Jason Gray, Robert Nachev, Parashkev Zhang, Hui Department of Computer Science Centre for Medical Image Computing University College London London United Kingdom Kagenova Limited Guildford United Kingdom Department of Brain Repair & Rehabilitation Institute of Neurology Ucl London United Kingdom

This paper demonstrates spherical convolutional neural networks (S-CNN) offer distinct advantages over conventional fully-connected networks (FCN) at estimating scalar parameters of tissue microstructure from diffusion MRI (dMRI). Such microstructure parameters are valuable for identifying pathology and quantifying its extent. However, current clinical practice commonly acquires dMRI data consisting of only 6 diffusion weighted images (DWIs), limiting the accuracy and precision of estimated microstructure indices. Machine learning (ML) has been proposed to address this challenge. However, existing ML-based methods are not robust to differing gradient schemes, nor are they rotation equivariant. Lack of robustness to gradient schemes requires a new network to be trained for each scheme, complicating the analysis of data from multiple sources. A possible consequence of the lack of rotational equivariance is that the training dataset must contain a diverse range of microstucture orientations. Here, we show spherical CNNs represent a compelling alternative that is robust to new gradient schemes as well as offering rotational equivariance. We show the latter can be leveraged to decrease the number of training datapoints required. © 2022, CC BY.

关键词： Magnetic resonance imaging

Real-Time Fault Detection and Diagnosis Method for Industrial Chemical Tennessee Eastman Process

学校读者我要写书评

暂无评论

Real-Time Fault Detection and Diagnosis Method for Industria...

International Conference on Control, Decision and Information Technologies (CoDIT)

作者： Khadija Attouri Majdi Mansouri Mansour Hajji Abdelmalek Kouadri Kais Bouzrara Hazem Nounou Institut Superieur des Sciences Appliques et de Technologie de Kasserine Kairouan University Tunisia The Department of Electrical and Computer Engineering Program Texas A&M University at Qatar Doha Qatar Signals and Systems Laboratory Institute of Electrical and Electronics Engineering University M Hamed Bougara of Boumerdes Boumerdes Algeria The Research Laboratory of Automation Signal Processing and Image National Engineering School of Monastir Tunisia

ISBN: (数字)9798350373974

ISBN: (纸本)9798350373981

The accurate detection and diagnosis of faults are critical for maintaining optimal operation and ensuring the reliability of industrial processes. Notably, the topic of online fault detection and diagnosis has recently presented a significant challenge. This work mainly deploys a neural network technique for the comprehensive detection and diagnosis of faults within the Tennessee Eastman Process (TEP) on a low-computational power system, the Raspberry Pi board. The devolved methodology showcases a remarkable level of accuracy (94.50%) in diagnosing the various TEP faults, affirming its robustness and effectiveness. To elevate the practical applicability of the proposed approach, a meticulous investigation into the implementation of the suggested approach on a Raspberry Pi 4 card was undertaken. The successful realization of this implementation not only highlights the adaptability of the approach but also paves the way for its seamless integration into practical industrial applications.

关键词： Accuracy Embedded systems Fault detection Neural networks Electrical fault detection Real-time systems Robustness Power system reliability Information technology Chemicals

A Fast Task Offloading Optimization Framework for IRS-Assisted Multi-Access Edge Computing System

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wu, Jianqiu Yu, Zhongyi Guo, Jianxiong Tang, Zhiqing Wang, Tian Jia, Weijia The Guangdong Key Lab of Ai and Multi-Modal Data Processing Department of Computer Science Bnuhkbu United International College 519087 Zhuhai China The Advanced Institute of Natural Sciences Beijing Normal University Zhuhai519087 China The Guangdong Key Lab of Ai and Multi-Modal Data Processing BNU-HKBU United International College Zhuhai519087 China

Terahertz communication networks and intelligent reflecting surfaces exhibit significant potential in advancing wireless networks, particularly within the domain of aerial-based multi-access edge computing systems. These technologies enable efficient offloading of computational tasks from user electronic devices to Unmanned Aerial Vehicles or local execution. For the generation of high-quality task-offloading allocations, conventional numerical optimization methods often struggle to solve challenging combinatorial optimization problems within the limited channel coherence time, thereby failing to respond quickly to dynamic changes in system conditions. To address this challenge, we propose a deep learning-based optimization framework called Iterative Order-Preserving policy Optimization (IOPO), which enables the generation of energy-efficient task-offloading decisions within milliseconds. Unlike exhaustive search methods, IOPO provides continuous updates to the offloading decisions without resorting to exhaustive search, resulting in accelerated convergence and reduced computational complexity, particularly when dealing with complex problems characterized by extensive solution spaces. Experimental results demonstrate that the proposed framework can generate energy-efficient task-offloading decisions within a very short time period, outperforming other benchmark methods. Copyright © 2023, The Authors. All rights reserved.

关键词： Iterative methods

Well-tuned simple nets excel on tabular datasets

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Kadra, Arlind Hutter, Frank Lindauer, Marius Grabocka, Josif Department of Computer Science University of Freiburg Germany Department of Computer Science University of Freiburg & Bosch Center for Artificial Intelligence Germany Institute for Information Processing Leibniz University Hannover Germany

Tabular datasets are the last "unconquered castle" for deep learning, with traditional ML methods like Gradient-Boosted Decision Trees still performing strongly even against recent specialized neural architectures. In this paper, we hypothesize that the key to boosting the performance of neural networks lies in rethinking the joint and simultaneous application of a large set of modern regularization techniques. As a result, we propose regularizing plain Multilayer Perceptron (MLP) networks by searching for the optimal combination/cocktail of 13 regularization techniques for each dataset using a joint optimization over the decision on which regularizers to apply and their subsidiary hyperparameters. We empirically assess the impact of these regularization cocktails for MLPs in a large-scale empirical study comprising 40 tabular datasets and demonstrate that (i) well-regularized plain MLPs significantly outperform recent state-of-the-art specialized neural network architectures, and (ii) they even outperform strong traditional ML methods, such as XGBoost. Copyright © 2021, The Authors. All rights reserved.

关键词： Network architecture

Uncertainty-aware Sampling for Long-tailed Semi-supervised Learning

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Yang, Kuo Li, Duo Hu, Menghan Zhai, Guangtao Yang, Xiaokang Zhang, Xiao-Ping The Shanghai Key Laboratory of Multidimensional Information Processing School of Communication and Electronic Engineering East China Normal University Shanghai200241 China The Kargobot of DiDi Shanghai201210 China The Institute of Image Communication and Network Engineering Shanghai Jiao Tong University Shanghai200240 China Tsinghua Berkeley Shenzhen Institute Shenzhen China The Department of Electrical Computer and Biomedical Engineering Toronto Metropolitan University ONM5B 2K3 Canada

For semi-supervised learning with imbalance classes, the long-tailed distribution of data will increase the model prediction bias toward dominant classes, undermining performance on less frequent classes. Existing methods also face challenges in ensuring the selection of sufficiently reliable pseudo-labels for model training and there is a lack of mechanisms to adjust the selection of more reliable pseudo-labels based on different training stages. To mitigate this issue, we introduce uncertainty into the modeling process for pseudo-label sampling, taking into account that the model performance on the tailed classes varies over different training stages. For example, at the early stage of model training, the limited predictive accuracy of model results in a higher rate of uncertain pseudo-labels. To counter this, we propose an Uncertainty-Aware Dynamic Threshold Selection (UDTS) approach. This approach allows the model to perceive the uncertainty of pseudo-labels at different training stages, thereby adaptively adjusting the selection thresholds for different classes. Compared to other methods such as the baseline method FixMatch, UDTS achieves an increase in accuracy of at least approximately 5.26%, 1.75%, 9.96%, and 1.28% on the natural scene image datasets CIFAR10-LT, CIFAR100-LT, STL-10-LT, and the medical image dataset TissueMNIST, respectively. The source code of UDTS is publicly available at: https://***/yangk/UDTS. Copyright © 2024, The Authors. All rights reserved.

关键词： Uncertainty analysis

Few-Shot Class-Incremental Learning with Prior Knowledge

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Jiang, Wenhao Li, Duo Hu, Menghan Zhai, Guangtao Yang, Xiaokang Zhang, Xiao-Ping Shanghai Key Laboratory of Multidimensional Information Processing School of Communication and Electronic Engineering East China Normal University Shanghai200241 China Kargobot of DiDi Shanghai201210 China Institute of Image Communication and Network Engineering Shanghai Jiao Tong University Shanghai200240 China Tsinghua Berkeley Shenzhen Institute Shenzhen China Department of Electrical Computer and Biomedical Engineering Toronto Metropolitan University ONM5B 2K3 Canada

To tackle the issues of catastrophic forgetting and overfitting in few-shot class-incremental learning (FSCIL), previous work has primarily concentrated on preserving the memory of old knowledge during the incremental phase. The role of pre-trained model in shaping the effectiveness of incremental learning is frequently underestimated in these studies. Therefore, to enhance the generalization ability of the pre-trained model, we propose Learning with Prior Knowledge (LwPK) by introducing nearly free prior knowledge from a few unlabeled data of subsequent incremental classes. We cluster unlabeled incremental class samples to produce pseudo-labels, then jointly train these with labeled base class samples, effectively allocating embedding space for both old and new class data. Experimental results indicate that LwPK effectively enhances the model resilience against catastrophic forgetting, with theoretical analysis based on empirical risk minimization and class distance measurement corroborating its operational principles. The source code of LwPK is publicly available at: https://***/StevenJ308/LwPK. Copyright © 2024, The Authors. All rights reserved.

关键词： Risk assessment

Minimal-time Deadbeat Consensus and Individual Disagreement Degree Prediction for High-order Linear Multi-agent Systems

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Hu, Fu-Long Zhang, Hai-Tao Xu, Bowen Hu, Zhe Ren, Wei School of Artificial Intelligence and Automation Key Laboratory of Image Processing and Intelligent Control The State Key Laboratory of Digital Manufacturing Equipment and Technology Huazhong University of Science and Technology Wuhan430074 China School of Artificial Intelligence Optics and Electronics Northwestern Polytechnical University Xi’an710072 China Department of Electrical and Computer Engineering University of California Riverside92521 United States

In this paper, a Hankel matrix-based fully distributed algorithm is proposed to address a minimal-time deadbeat consensus prediction problem for discrete-time high-order multi-agent systems (MASs). Therein, each agent can predict the consensus value with the minimum number of observable historical outputs of its own. Accordingly, compared to most existing algorithms only yielding asymptotic convergence, the present method can attain deadbeat consensus instead. Moreover, based on the consensus value prediction, instant individual disagreement degree value of MASs can be calculated in advance as well. Sufficient conditions are derived to guarantee both the minimal-time deadbeat consensus and the instant individual disagreement degree prediction. Finally, both the effectiveness and superiority of the proposed deadbeat consensus algorithm are substantiated by numerical simulations. Copyright © 2023, The Authors. All rights reserved.

关键词： Multi agent systems