检索结果-内蒙古大学图书馆

Flare Forecast Model Based on DS-SMOTE and SVM with Optimized Regular Term

Research in Astronomy and Astrophysics 2023年第6期23卷 38-46页

作者： Jie Wan Jun-Feng Fu Ren-Qing Wen Ke Han Meng-Yao Yu Peng E School of Electrical Engineering and Automation Harbin Institute of TechnologyHarbin 150001China Laboratory for Space Environment and Physical Sciences Harbin Institute of TechnologyHarbin 150001China School of Computer and Information Engineering Harbin University of CommerceHarbin 150028China

The research of flare forecast based on the machine learning algorithm is an important content of space *** order to improve the reliability of the data-driven model and weaken the impact of imbalanced data set on its forecast performance,we proposes a resampling method suitable for flare forecasting and a Particle Swarm Optimization(PSO)-based Support Vector Machine(SVM)regular term optimization *** the problem of intra-class imbalance and inter-class imbalance in flare samples,we adopt the density clustering method combined with the Synthetic Minority Over-sampling Technique(SMOTE)oversampling method,and performs the interpolation operation based on Euclidean distance on the basis of analyzing the clustering space in the minority *** the same time,for the problem that the objective function used for strong classification in SVM cannot adapt to the sample noise,In this research,on the basis of adding regularization parameters,the PSO algorithm is used to optimize the hyperparameters,which can maximize the performance of the ***,through a comprehensive comparison test,it is proved that the method designed can be well applied to the flare forecast problem,and the effectiveness of the method is proved.

关键词： Sun:flares Sun:magnetic fields Sun:X-rays gamma-rays (Sun:)sunspots

来源：评论

学校读者我要写书评

暂无评论

MemDefense: Defending against Membership Inference Attacks in IoT-based Federated Learning via Pruning Perturbations

引用

IEEE Transactions on Big Data 2024年 1-13页

作者： Shen, Meng Meng, Jin Xu, Ke Yu, Shui Zhu, Liehuang School of Cyberspace Science and Technology Beijing Institute of Technology Beijing China Department of Computer Science Tsinghua University Beijing China School of Computer Science University of Technology Sydney Sydney NSW Australia

Depending on large-scale devices, the Internet of Things (IoT) provides massive data support for resource sharing and intelligent decision, but privacy risks also increase. As a popular distributed learning framework, Federated Learning (FL) is widely used because it does not need to share raw data while only parameters to collaboratively train models. However, Federated Learning is not spared by some emerging attacks, e.g., membership inference attack. Therefore, for IoT devices with limited resources, it is challenging to design a defense scheme against the membership inference attack ensuring high model utility, strong membership privacy and acceptable time efficiency. In this paper, we propose MemDefense, a lightweight defense mechanism to prevent membership inference attack from local models and global models in IoT-based FL, while maintaining high model utility. MemDefense adds crafted pruning perturbations to local models at each round of FL by deploying two key components, i.e., parameter filter and noise generator. Specifically, the parameter filter selects the apposite model parameters which have little impact on the model test accuracy and contribute more to member inference attacks. Then, the noise generator is used to find the pruning noise that can reduce the attack accuracy while keeping high model accuracy, protecting each participant's membership privacy. We comprehensively evaluate MemDefense with different deep learning models and multiple benchmark datasets. The experimental results show that lowcost MemDefense drastically reduces the attack accuracy within limited drop of classification accuracy, meeting the requirements for model utility, membership privacy and time efficiency. IEEE

关键词： Internet of things

来源：评论

学校读者我要写书评

暂无评论

Cryptography with optimal deep learning-based authentication scheme for preserving anonymity in telecare medical information system

引用

Multimedia Tools and Applications 2025年第10期84卷 8311-8330页

作者： Sheik, Syed Amma Durai, Saleem School of Computer Science and Engineering Vellore Institute of Technology Tamil Nadu India

The Telecare Medicine Information System (TMIS) revolutionizes healthcare delivery by integrating medical equipment and sensors, facilitating proactive and cost-effective services. Accessible online, TMIS empowers patients and practitioners to manage medical data efficiently, ensuring timely care provision and streamlined record maintenance. Recognizing security challenges inherent in this interconnected healthcare landscape, researchers have devised a two-factor authentication solution, increasingly leveraging password-based authentication via smart cards. This article presents the development of a tailored remote authentication system, the Deep Learning-Based Online Threat Detection with Key Agreement (DLOTDKA-MAP), designed specifically for TMIS. Comprising two critical phases, the DLOTDKA-MAP framework employs an Arithmetic Optimization Algorithm (AOA) and Deep Belief Network (DBN) model in real-time online threat detection, ensuring swift identification and mitigation of potential security threats. The subsequent phase employs Elliptic Curve Cryptography (ECC) with Schnorr's signature to establish secure data transmission among collaborating entities, preemptively addressing security breaches and ensuring mutual authentication. A comprehensive experimental study validates the presented DLOTDKA-MAP technique, showcasing its superiority over existing approaches and underscoring its role in advancing TMIS security infrastructure. Achieving remarkable performance values, includes an accuracy, precision, recall, and F1 score of 98.57%, this research significantly contributes to safeguarding the confidentiality and integrity of medical data within TMIS, particularly in a federated cloud environment. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Authentication

来源：评论

学校读者我要写书评

暂无评论

Efficient breast cancer detection using neural networks and explainable artificial intelligence

引用

Neural Computing and Applications 2025年第5期37卷 3759-3776页

作者： Murugan, Tamilarasi Kathirvel Karthikeyan, Pritikaa Sekar, Pavithra School of Computer Science Engineering Vellore Institute of Technology Tamilnadu Chennai India

The growing dependence on deep learning models for medical diagnosis underscores the critical need for robust interpretability and transparency to instill trust and ensure responsible usage. This study investigates the efficacy of various explainable artificial intelligence (XAI) techniques in comprehending deep learning models utilized for breast cancer classification from down sampled histopathology images. A comparative assessment of multiple convolutional neural network (CNN) architectures, encompassing standard CNNs, ResNet, VGG-16, and VGG-19, on down sampled images was conducted. The primary goal is to pinpoint the model exhibiting the highest accuracy and subsequently employ three prominent XAI methods—LIME, SHAP, and Saliency Maps—to get insights into the top-performing model. This study identifies VGG-19 as the best-performing model with an accuracy of 92.59% and demonstrates that among various XAI techniques, LIME provides the most accurate and clinically relevant explanations for breast cancer classification from down sampled histopathology images. These findings, validated by medical professionals, enhance the interpretability and reliability of deep learning models in clinical settings, promoting their responsible integration into healthcare practices. This validation was further corroborated through consultation with medical professionals, including doctors specializing in breast cancer diagnosis. This research endeavors to deepen the understanding of the model’s rationale and instill confidence in its outputs. The outcomes of this study hold significant promise in elevating the interpretability and reliability of deep learning models tailored for breast cancer diagnosis, thus facilitating their responsible integration into clinical settings. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Parametric control of quasi-linear second-order systems with partitioned eigenstructure assignment by output feedback

引用

science China(Information sciences) 2023年第4期66卷 228-243页

作者： Weizhen LIU Guangren DUAN Dake GU Center for Control Theory and Guidance Technology Harbin Institute of Technology Center for Control Science and Technology Southern University of Science and Technology School of Automation Engineering Northeast Electric Power University

In this paper, a parametric design approach for stabilizing a quasi-linear second-order system with partitioned eigenstructure assignment(PESA) is investigated through output feedback control. The PESA approach is established by partitioning the desired eigenvalue matrix into two parts to separate the associated right and left eigenvectors into a subset of the generalized eigenvectors simultaneously. A parametric controller is established by solving two second-order generalized Sylvester matrix equations, and a certain form with the desired eigenstructure can be derived with the established quasi-linear output feedback controller. Unlike the prevailing approach that assigns the entire set of generalized eigenvectors, which is difficult to satisfy a large number of complicated constraints in practical systems by the normalized pair of right and left eigenvector matrices, a subset of the generalized eigenvectors is considered. In addition, the proposed PESA approach provides less computational load and is easy to use. A numerical example and application in spacecraft rendezvous are provided to verify the numerical economy and high efficiency of the proposed approach.

关键词： quasi-linear second-order systems partitioned eigenstructure assignment parametric approach Sylvester matrix equation output feedback

来源：评论

学校读者我要写书评

暂无评论

MTST: A Multi-Task Scheduling Transformer Accelerator for Edge Computing 13

MTST: A Multi-Task Scheduling Transformer Accelerator for Ed...

引用

13th IEEE Global Conference on Consumer Electronic, GCCE 2024

作者： Yue, Zongcheng Yan, Dongwei Wu, Ran Ma, Longyu Sham, Chiu-Wing The University of Auckland School of Computer Science New Zealand Harbin Institute of Technology School of Computer Science China

ISBN: (纸本)9798350355079

Transformer is pivotal in Large Language Models (LLMs), enabling superior performance in language tasks. However, the abundance of parameters poses a challenge for deploying Transformer on resource-constrained edge devices for edge computing, such as Field-Programmable Gate Arrays (FPGAs). To overcome this limitation, we present an FPGA-based architecture that enables high-performance deployment of Transformer for edge computing, incorporating a unique multitask ping-pong scheme. To strike a balance between computing efficiency and logic resource consumption, we adopt a systolic array as a reusable processing engine (PE), employing meticulous space exploration techniques to determine the optimal quantity and dimensions of PEs within a given resource budget for our architecture. We implement the architecture on ZCU 102 FPGA platform, the experimental results illustrate that the proposed accelerator achieves state-of-the-art performance while consuming fewer resources. © 2024 IEEE.

关键词： Budget control

来源：评论

学校读者我要写书评

暂无评论

GDMNet: A Unified Multi-Task Network for Panoptic Driving Perception

引用

computers, Materials & Continua 2024年第8期80卷 2963-2978页

作者： Yunxiang Liu Haili Ma Jianlin Zhu Qiangbo Zhang School of Computer Science and Information Engineering Shanghai Institute of TechnologyShanghai201418China

To enhance the efficiency and accuracy of environmental perception for autonomous vehicles,we propose GDMNet,a unified multi-task perception network for autonomous driving,capable of performing drivable area segmentation,lane detection,and traffic object ***,in the encoding stage,features are extracted,and Generalized Efficient Layer Aggregation Network(GELAN)is utilized to enhance feature extraction and gradient ***,in the decoding stage,specialized detection heads are designed;the drivable area segmentation head employs DySample to expand feature maps,the lane detection head merges early-stage features and processes the output through the Focal Modulation Network(FMN).Lastly,the Minimum Point Distance IoU(MPDIoU)loss function is employed to compute the matching degree between traffic object detection boxes and predicted boxes,facilitating model training *** results on the BDD100K dataset demonstrate that the proposed network achieves a drivable area segmentation mean intersection over union(mIoU)of 92.2%,lane detection accuracy and intersection over union(IoU)of 75.3%and 26.4%,respectively,and traffic object detection recall and mAP of 89.7%and 78.2%,*** detection performance surpasses that of other single-task or multi-task algorithm models.

关键词： Autonomous driving multitask learning drivable area segmentation lane detection vehicle detection

来源：评论

学校读者我要写书评

暂无评论

High-Resolution Image Harmonization with Adaptive-Interval Color Transformation 38

High-Resolution Image Harmonization with Adaptive-Interval C...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Meng, Quanling Liu, Qinglin Li, Zonglin Lan, Xiangyuan Zhang, Shengping Nie, Liqiang School of Computer Science and Technology Harbin Institute of Technology China Peng Cheng Laboratory China

Existing high-resolution image harmonization methods typically rely on global color adjustments or the upsampling of parameter maps. However, these methods ignore local variations, leading to inharmonious appearances. To address this problem, we propose an Adaptive-Interval Color Transformation method (AICT), which predicts pixel-wise color transformations and adaptively adjusts the sampling interval to model local non-linearities of the color transformation at high resolution. Specifically, a parameter network is first designed to generate multiple position-dependent 3-dimensional lookup tables (3D LUTs), which use the color and position of each pixel to perform pixel-wise color transformations. Then, to enhance local variations adaptively, we separate a color transform into a cascade of sub-transformations using two 3D LUTs to achieve the non-uniform sampling intervals of the color transform. Finally, a global consistent weight learning method is proposed to predict an image-level weight for each color transform, utilizing global information to enhance the overall harmony. Extensive experiments demonstrate that our AICT achieves state-of-the-art performance with a lightweight architecture. The code is available at https://***/aipixel/AICT. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Texture evolution prediction of 2219 aluminum alloy sheet under hydro-bulging using cross-scale numerical modeling

引用

Journal of Materials science & technology 2023年第18期149卷 190-204页

作者： Yanbo Pei Yonggang Hao Jie Zhao Jiantong Yang Bugang Teng School of Materials Science and Engineering Harbin Institute of TechnologyHarbin 150001China National Key Laboratory for Precision Hot Processing of Metals Harbin Institute of TechnologyHarbin 150001China Department of Materials Science and Engineering China University of Mining&Technology(Beijing)Beijing 100083China

A simultaneous prediction of macroscopic deformation and microstructure evolution is critical for un-derstanding the deformation mechanism of *** this work,the hydro-bulging process of 2219 aluminum alloy sheet was investigated using cross-scale numerical modeling,in which the macroscopic finite element method(FEM)and crystal plasticity finite element method(CPFEM)were *** calculated texture evolution exhibits good agreement with the experimental results,and the stress er-ror between the two scales is generally *** effects of different strain states on texture evolution and slip mode are further *** the strain ratioηincreases,the volume fractions of the initial Rotated Copper texture component andγ-Fiber texture component decrease significantly,which tend to be stabilized at P texture *** initial Rotated Cube texture component is inclined to rotate towards the Cube texture component,while the volume fraction of this orientation is relatively *** lower strain ratio can considerably enhance the activity of more equivalent slip systems,promoting a more uniform strain distribution over *** difficulty of grain deformation changes as the lat-tice *** grain with easy-to-deform orientation can gradually rotate to a stable orientation during plastic deformation,which has a lower Schmid factor.

关键词： Cross-scale modeling Crystal plasticity Texture evolution Aluminum alloy Hydro-bulging forming

来源：评论

学校读者我要写书评

暂无评论

Learning Visual Representation of Underwater Acoustic Imagery Using Transformer-Based Style Transfer Method 23

Learning Visual Representation of Underwater Acoustic Imager...

引用

15th International Conference on Digital Image Processing, ICDIP 2023

作者： Zhou, Xiaoteng Yu, Changli Yuan, Shihao Yuan, Xin Yu, Hangchi Luo, Citong School of Ocean Engineering Harbin Institute of Technology China School of Computer Science and Technology Harbin Institute of Technology China

ISBN: (纸本)9798400708237

Underwater automatic target recognition (UATR) has been a challenging research topic in ocean engineering. Although deep learning brings opportunities for target recognition on land and in the air, underwater target recognition techniques based on deep learning have lagged due to sensor performance and the size of trainable data. This letter proposed a framework for learning the visual representation of underwater acoustic imageries, which takes a transformer-based style transfer model as the main body. It could replace the low-level texture features of optical images with the visual features of underwater acoustic imageries while preserving their raw high-level semantic content. The proposed framework could fully use the rich optical image dataset to generate a pseudo-acoustic image dataset and use it as the initial sample to train the underwater acoustic target recognition model. The experiments select the dual-frequency identification sonar (DIDSON) and adaptive resolution imaging sonar (ARIS) as the underwater acoustic data source and also take fish, the most common marine creature, as the research subject. Experimental results show that the proposed method could generate high-quality and high-fidelity pseudo-acoustic samples, achieve the purpose of acoustic data augmentation and provide support for the underwater acoustic-optical images domain transfer research. © 2023 ACM.

关键词： Underwater acoustics

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：