检索结果-内蒙古大学图书馆

Journal of Geophysical Research: Machine Learning and Computation 2024年第4期1卷

作者： Donsub Rim Sanah Suri Sanghyun Hong Kookjin Lee Randall J. LeVeque Department of Mathematics Washington University in St. Louis St. Louis MO USA School of Electrical Engineering and Computer Science Oregon State University Corvallis OR USA School of Computing Informatics and Decision Systems Engineering Arizona State University Tempe AZ USA Department of Applied Mathematics University of Washington Seattle WA USA

Neural networks (NNs) enable precise modeling of complicated geophysical phenomena but can be sensitive to small input changes. In this work, we present a new method for analyzing this instability in NNs. We focus our analysis on adversarial examples, test-time inputs with carefully crafted human-imperceptible perturbations that expose the worst-case instability in a model's predictions. Our stability analysis is based on a low-rank expansion of NNs on a fixed input, and we apply our analysis to a NN model for tsunami early warning which takes geodetic measurements as the input and forecasts tsunami waveforms. The result is an improved description of local stability that explains adversarial examples generated by a standard gradient-based algorithm, and allows the generation of other comparable examples. Our analysis can predict whether noise in the geodetic input will produce an unstable output, and identifies a potential approach to filtering the input that enable more robust forecasting. Neural networks (NNs) suffer from input instabilities, making them difficult to deploy in critical applications like natural hazard warning We derive a low-rank expansion for NNs, revealing an input-dependent low-rank matrix that describes local input stability We illustrate that the instabilities are closely related to the singular vectors derived from the low-rank matrix Deep learning models trained on empirical or simulated data can yield accurate predictions in real time. Consequently, their use has expanded to predicting complex physical phenomena. For example, a convolutional neural network can predict tsunami waveforms accurately using only geodetic measurements of the earthquake, thereby potentially informing early warning systems. However, indiscernible perturbations in the measurement data can incur instabilities in neural networks (NNs). Should the early warning systems rely on such models, the ramifications of unstable predictions are of great concern for safety and s

关键词： stability adversarial examples neural networks tsunami early warning low-rank expansions

来源：评论

学校读者我要写书评

暂无评论

A new evolutionary algorithm: Learner performance based behavior algorithm

arXiv

引用

arXiv 2020年

作者： Rahman, Chnoor M. Rashid, Tarik A. Applied Computer Department College of Medicals and Applied Sciences Charmo University SulaimanyKRG Iraq Technical College of Informatics Sulaimany Polytechnic University SulaimanyKRG Iraq Computer Science and Engineering Department University of Kurdistan Hewler ErbilKRG Iraq

A novel evolutionary algorithm called learner performance based behavior algorithm (LPB) is proposed in this article. The basic inspiration of LPB originates from the process of accepting graduated learners from high school in different departments at university. In addition, the changes those learners should do in their studying behaviors to improve their study level at university. The most important stages of optimization;exploitation and exploration are outlined by designing the process of accepting graduated learners from high school to university and the procedure of improving the learner’s studying behavior at university to improve the level of their study. To show the accuracy of the proposed algorithm, it is evaluated against a number of test functions, such as traditional benchmark functions, CEC-C06 2019 test functions, and a real-world case study problem. The results of the proposed algorithm are then compared to the DA, GA, and PSO. The proposed algorithm produced superior results in most of the cases and comparative in some others. It is proved that the algorithm has a great ability to deal with the large optimization problems comparing to the DA, GA, and PSO. The overall results proved the ability of LPB in improving the initial population and converging towards the global optima. Moreover, the results of the proposed work are proved statistically. Copyright © 2020, The Authors. All rights reserved.

关键词： Genetic algorithms

来源：评论

学校读者我要写书评

暂无评论

Pruning neural networks without any data by iteratively conserving synaptic flow

arXiv

引用

arXiv 2020年

作者： Tanaka, Hidenori Yamins, Daniel L.K. Kunin, Daniel Ganguli, Surya Physics & Informatics Laboratories NTT Reserach Inc Department of Applied Physics Stanford University Department of Psychology Department of Computer Science Stanford University Institute for Computational and Mathematical Engineering Stanford University Department of Applied Physics Stanford University

Pruning the parameters of deep neural networks has generated intense interest due to potential savings in time, memory and energy both during training and at test time. Recent works have identified, through an expensive sequence of training and pruning cycles, the existence of winning lottery tickets or sparse trainable subnetworks at initialization. This raises a foundational question: can we identify highly sparse trainable subnetworks at initialization, without ever training, or indeed without ever looking at the data? We provide an affirmative answer to this question through theory driven algorithm design. We first mathematically formulate and experimentally verify a conservation law that explains why existing gradient-based pruning algorithms at initialization suffer from layer-collapse, the premature pruning of an entire layer rendering a network untrainable. This theory also elucidates how layer-collapse can be entirely avoided, motivating a novel pruning algorithm Iterative Synaptic Flow Pruning (SynFlow). This algorithm can be interpreted as preserving the total flow of synaptic strengths through the network at initialization subject to a sparsity constraint. Notably, this algorithm makes no reference to the training data and consistently outperforms existing state-of-the-art pruning algorithms at initialization over a range of models (VGG and ResNet), datasets (CIFAR-10/100 and Tiny ImageNet), and sparsity constraints (up to 99.9 percent). Thus our data-agnostic pruning algorithm challenges the existing paradigm that data must be used to quantify which synapses are important. Copyright © 2020, The Authors. All rights reserved.

关键词： Iterative methods

来源：评论

学校读者我要写书评

暂无评论

The rank of contextuality

arXiv

引用

arXiv 2022年

作者： Horodecki, Karol Zhou, Jingfang Stankiewicz, Maciej Salazar, Roberto Horodecki, Pawel Raussendorf, Robert Horodecki, Ryszard Ramanathan, Ravishankar Tyhurst, Emily Institute of Informatics National Quantum Information Centre Faculty of Mathematics Physics and Informatics University of Gdańsk Wita Stwosza 57 Gdańsk80-308 Poland International Centre for Theory of Quantum Technologies University of Gdańsk Wita Stwosza 63 Gdańsk80-308 Poland Department of Applied Physics Graduate School of Engineering University of Tokyo 7-3-1 Hongo Bunkyo-ku Tokyo113-8656 Japan Faculty of Physics Astronomy and Applied Computer Science Jagiellonian University Kraków30-348 Poland Faculty of Applied Physics and Mathematics National Quantum Information Centre Gdańsk University of Technology Gabriela Narutowicza 11/12 Gdańsk80-233 Poland Department of Physics and Astronomy University of British Columbia VancouverBCV6T 1Z1 Canada Department of Computer Science The University of Hong Kong Pokfulam Road Hong Kong Department of Physics University of Toronto TorontoON Canada

Quantum contextuality is one of the most recognized resources in quantum communication and computing scenarios. We provide a new quantifier of this resource, the rank of contextuality (RC). We define RC as the minimum number of non-contextual behaviors that are needed to simulate a contextual behavior. We show that the logarithm of RC is a natural contextuality measure satisfying several properties considered in the spirit of the resource-theoretic approach. The properties include faithfulness, monotonicity, and additivity under tensor product. We also give examples of how to construct contextual behaviors with an arbitrary value of RC exhibiting a natural connection between this quantifier and the arboricity of an underlying hypergraph. We also discuss exemplary areas of research in which the new measure appears as a natural quantifier. Copyright © 2022, The Authors. All rights reserved.

关键词： Quantum communication

来源：评论

学校读者我要写书评

暂无评论

Corrections to “Statistical Insights Into Machine Learning-Based Box Models for Pregnancy Care and Maternal Mortality Reduction: A Literature Survey”

引用

IEEE Access 2024年 12卷 136435-136435页

作者： Issac Neha Margret K. Rajakumar K. V. Arulalan S. Manikandan Valentina E. Balas School of Computer Science and Engineering Vellore Institute of Technology Vellore Tamil Nadu India A. A. Child Care Clinic Vellore Tamil Nadu India DRDO Bengaluru Karnataka India Department of Automation and Applied Informatics Aurel Vlaicu University of Arad Arad Romania

Presents corrections to the paper, Corrections to “Statistical Insights Into Machine Learning-Based Box Models for Pregnancy Care and Maternal Mortality Reduction: A Literature Survey”.

关键词： Pregnancy

来源：评论

学校读者我要写书评

暂无评论

Deep Reinforcement Learning-based Obstacle Avoidance for Robot Movement in Warehouse Environments

Deep Reinforcement Learning-based Obstacle Avoidance for Rob...

引用

IEEE International Conference on Civil Aviation Safety and Information Technology (ICCASIT)

作者： Keqin Li Jiajing Chen Dezhi Yu Tao Dajun Xinyu Qiu Jieting Lian Ryan Ji Shengyuan Zhang Zhenyu Wan Baiwei Sun Bo Hong Fanghao Ni Jiaqi Han Department of Computer Science AMA University Quezon Philippines Courant Institute New York University 251 Mercer St New York NY USA School of Information University of California Berkeley University Avenue and Oxford St Berkeley CA United States Master of Science in Mechanical Engineering School of Engineering Carnegie Mellon University 5000 Forbes Avenue Pittsburgh PA USA College of Engineering Northeastern University Seattle USA Master of Science in Database Technology School of Professional Studies New York University 50 West 4th Street New York NY USA College of Engineering Cornell University 616 Thurston Ave Ithaca NY US Master of Professional Science in Applied Statistics-Data Science College of Computing and Information Science Cornell University 616 Thurston Ave Ithaca NY USA Independent Researcher Beijing China Master of Computer Science Donald Bren School of Information and Computer Sciences University of California Irvine 510 Aldrich Hall Irvine CA USA School of Informatics Computing and Cyber Systems Northern Arizona University Flagstaff AZ USA Shanghai Institute of Technology Shanghai China

ISBN: (数字)9798350389418

ISBN: (纸本)9798350389425

This paper proposes a deep reinforcement learning based on the warehouse environment, the mobile robot obstacle avoidance Algorithm. Firstly, the value function network is improved based on the pedestrian interaction. We can learn to obtain the relative importance of the current state and the historical trajectory state as well as the joint impact on the robot’s obstacle avoidance strategy. Secondly, the reward function of reinforcement learning is designed based on the spatial behaviour of pedestrians, and the robot is punished for the state where the angle changes too much, so as to achieve the requirement of comfortable obstacle avoidance. Finally, the feasibility and effectiveness of the deep reinforcement learning-based mobile robot obstacle avoidance algorithm in the warehouse environment in the complex environment of the warehouse are verified through simulation experiments.

关键词： Legged locomotion Pedestrians Heuristic algorithms Robot sensing systems Feature extraction Trajectory Safety Data mining Collision avoidance Information technology

来源：评论

学校读者我要写书评

暂无评论

Pruning neural networks without any data by iteratively conserving synaptic flow 20

Pruning neural networks without any data by iteratively cons...

引用

Proceedings of the 34th International Conference on Neural Information Processing Systems

作者： Hidenori Tanaka Daniel Kunin Daniel L. K. Yamins Surya Ganguli Physics & Informatics Laboratories NTT Research Inc. Department of Applied Physics Stanford University Institute for Computational and Mathematical Engineering Stanford University Department of Psychology Department of Computer Science Stanford University Department of Applied Physics Stanford University

ISBN: (纸本)9781713829546

Pruning the parameters of deep neural networks has generated intense interest due to potential savings in time, memory and energy both during training and at test time. Recent works have identified, through an expensive sequence of training and pruning cycles, the existence of winning lottery tickets or sparse trainable subnetworks at initialization. This raises a foundational question: can we identify highly sparse trainable subnetworks at initialization, without ever training, or indeed without ever looking at the data? We provide an affirmative answer to this question through theory driven algorithm design. We first mathematically formulate and experimentally verify a conservation law that explains why existing gradient-based pruning algorithms at initialization suffer from layer-collapse, the premature pruning of an entire layer rendering a network untrainable. This theory also elucidates how layer-collapse can be entirely avoided, motivating a novel pruning algorithm Iterative Synaptic Flow Pruning (SynFlow). This algorithm can be interpreted as preserving the total flow of synaptic strengths through the network at initialization subject to a sparsity constraint. Notably, this algorithm makes no reference to the training data and consistently competes with or outperforms existing state-of-the-art pruning algorithms at initialization over a range of models (VGG and ResNet), datasets (CIFAR-10/100 and Tiny ImageNet), and sparsity constraints (up to 99.99 percent). Thus our data-agnostic pruning algorithm challenges the existing paradigm that, at initialization, data must be used to quantify which synapses are important.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Spoofing detection on hand images using quality assessment

arXiv

引用

arXiv 2021年

作者： Bera, Asish Dey, Ratnadeep Bhattacharjee, Debotosh Nasipuri, Mita Shum, Hubert P.H. Department of Computer Science Edge Hill University Lancashire OrmskirkL39 4QP United Kingdom Department of Computer Science and Engineering Jadavpur University Kolkata700032 India Center for Basic and Applied Science Faculty of informatics and management University of Hradec Kralove Rokitanskeho 62 Hradec Kralove500 03 Czech Republic Department of Computer Science Durham University Stockton Road DurhamDH1 3LE United Kingdom

Recent research on biometrics focuses on achieving a high success rate of authentication and addressing the concern of various spoofing attacks. Although hand geometry recognition provides adequate security over unauthorized access, it is susceptible to presentation attack. This paper presents an anti-spoofing method toward hand biometrics. A presentation attack detection approach is addressed by assessing the visual quality of genuine and fake hand images. A threshold-based gradient magnitude similarity quality metric is proposed to discriminate between the real and spoofed hand samples. The visual hand images of 255 subjects from the Bogazici University hand database are considered as original samples. Correspondingly, from each genuine sample, we acquire a forged image using a Canon EOS 700D camera. Such fake hand images with natural degradation are considered for electronic screen display based spoofing attack detection. Furthermore, we create another fake hand dataset with artificial degradation by introducing additional Gaussian blur, salt and pepper, and speckle noises to original images. Ten quality metrics are measured from each sample for classification between original and fake hand image. The classification experiments are performed using the k-nearest neighbors, random forest, and support vector machine classifiers, as well as deep convolutional neural networks. The proposed gradient similarity-based quality metric achieves 1.5% average classification error using the k-nearest neighbors and random forest classifiers. An average classification error of 2.5% is obtained using the baseline evaluation with the MobileNetV2 deep network for discriminating original and different types of fake hand samples. Copyright © 2021, The Authors. All rights reserved.

关键词： k-nearest neighbors

来源：评论

学校读者我要写书评

暂无评论

SmartCHANGE: AI-based long-term health risk evaluation for driving behaviour change strategies in children and youth

SmartCHANGE: AI-based long-term health risk evaluation for d...

引用

International Conference on applied Mathematics & computer Science (ICAMCS)

作者： Nina Reščič Janna Alberts Teatske M. Altenburg Mai J. M. Chinapaw Antonio De Nigro Dario Fenoglio Martin Gjoreski Anton Gradišek Gregor Jurak Athanasios Kiourtis Dimosthenis Kyriazis Marc Langheinrich Elena Mancuso Argyro Mavrogiorgou Mykola Pechenizkiy Roberto Pratola José Ribeiro Maroje Sorić Fawad Taj Tuija H. Tammelin Martijn Vastenburg Anna Vilanova Tanja G. M. Vrijkotte Mitja Luštrek Department of Intelligent Systems Jožef Stefan Institute Ljubljana Slovenia Research & Development ConnectedCare Nijmegen The Netherlands Department of Public and Occupational Health Amsterdam UMC Amsterdam Public Health Research Institute Vrije Universiteit Amsterdam Amsterdam The Netherlands Engineering Ingegneria Informatica S.p.A. Rome Italy Faculty of Informatics Università Della Svizzera Italiana Lugano Switzerland Faculty of Sports University of Ljubljana Ljubljana Slovenia Department of Digital Systems University of Piraeus Piraeus Greece Department of Mathematics and Computer Science Eindhoven University of Technology Eindhoven Netherlands Faculty of Sport University of Porto Porto Portugal Likes School of Health and Social Studies Jamk University of Applied Sciences Jyväskylä Finland

The SmartCHANGE project, a Horizon-Europe Research & Innovation project that has been ongoing since May 2023 and is scheduled to conclude in April 2027, aims to develop AI-driven decision-support tools to identify and reduce the long-term risk of non-communicable chronic diseases (NCDs) in children and youth. NCDs have become a significant public health burden in developed countries, with common risk factors including obesity, low physical fitness, and unhealthy lifestyle habits. Childhood and adolescence are critical periods for establishing healthy habits, and early intervention can prevent or delay the onset of NCDs later in life. However, current tools for identifying high-risk individuals mainly focus on adults, leading to missed opportunities for early detection. The SmartCHANGE project aims to address this gap by creating trustworthy AI tools that accurately assess risk factors in children and youth and promote optimised strategies for risk reduction. The paper outlines the plan for the project from the project’s proposed methodologies, including dataset utilisation, data cleaning, risk prediction models (basic and complex), and the implementation of federated learning. Additionally, it discusses user experience aspects such as participatory design, suggested applications, and the evaluation of project outcomes.

关键词：

来源：评论

学校读者我要写书评

暂无评论

C3BAM-XAI: Convolutional Block Attention Module Enhanced Explainable Artificial Intelligence-Based Parkinson’s Disease Stage Classification

引用

Cognitive Computation 2025年第3期17卷 1-19页

作者： Abbas, Muhammad John Khan, Muhammad Attique Hamza, Ameer Alsenan, Shrooq Rehman, Aleesha Baili, Jamel Zhang, Yudong Department of Computer Science HITEC University Taxila47080 Pakistan Center of Artificial Intelligence College of Computer Engineering and Science Prince Mohammad Bin Fahd University Al-Khobar Saudi Arabia Centre of Real Time Computer Systems Kaunas University of Technology Kaunas51368 Lithuania Information Systems Department College of Computer and Information Sciences Princess Nourah Bint Abdulrahman University P.O. Box 84428 Riyadh11671 Saudi Arabia Pak-Austria Fachhochschule Institute of Applied Sciences and Technology Haripur Mang Pakistan Department of Computer Engineering College of Computer Science King Khalid University Abha61413 Saudi Arabia Department of Informatics University of Leicester Leicester United Kingdom

Parkinson’s disease (PD) is a progressive neurodegenerative disorder that causes significant impairment in neurons, physiological structures, and behavior of people. However, these changes are very subtle in the early stages of PD, making diagnosis and treatment challenging. To overcome such challenges, we proposed a novel convolutional neural network (CNN) architecture called C3BAM-XAI that is based on the convolutional block attention module (CBAM) enhanced CNN and explainable artificial intelligence (XAI). In the proposed architecture, the data imbalance issue was resolved at the initial stage by employing augmentation techniques for the training data. After that, the proposed CNN architecture is designed with CBAM. The CBAM architecture is based on two sub-modules—the channel attention module and the spatial attention module. The input of the initial convolutional layer is passed to the channel attention module, which is further passed to the spatial attention module for more critical information extraction. For smooth training, hyperparameters are selected using Nadam Optimization (NO) instead of manual assignment. In the testing phase, features are extracted from the dense layer and passed to several neural network classifiers. In addition to this, the designed model is interpreted using the explainable AI technique to see the predicted label and insight decision features. The experimental process was performed on the publically available PD Kaggle dataset and obtained an improved accuracy of 93.33%. In addition, several ablation studies were conducted, and it is concluded that the CBAM not only improved accuracy but also made this model a reliable tool for clinical diagnosis, both in terms of accuracy and interpretability. The results thus open new avenues toward applying deep learning models practically for applications in the medical domain and early-stage detection of Parkinson’s disease, thereby helping to improve patient outcomes. © The Author(s), under

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：