检索结果-内蒙古大学图书馆

Toward actionable testing of deep learning models

science China(Information sciences) 2023年第7期66卷 300-302页

作者： Yingfei XIONG Yongqiang TIAN Yepang LIU Shing-Chi CHEUNG Key Laboratory of High Confidence Software Technologies(Peking University) Ministry of Education School of Computer Science Peking University Department of Computer Science and Engineering The Hong Kong University of Science and Technology School of Computer Science University of Waterloo Research Institute of Trustworthy Autonomous Systems Southern University of Science and Technology Department of Computer Science and Engineering Southern University of Science and Technology

Deep learning has become an important computational paradigm in our daily lives with a wide range of applications,from authentication using facial recognition to autonomous driving in smart vehicles. The quality of the deep learning models, i.e., neural architectures with parameters trained over a dataset, is crucial to our daily living and economy.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Improving coronary heart disease prediction with real-life dataset: a stacked generalization framework with maximum clinical attributes and SMOTE balancing for imbalanced data

引用

Multimedia Tools and Applications 2024年第37期83卷 85139-85168页

作者： Dubey, Madhuri Tembhurne, Jitendra Makhijani, Richa Department of Computer Science and Engineering Indian Institute of Information Technology Maharashtra Nagpur441108 India

Heart disease increases the strain on the heart by reducing its ability to pump blood throughout the body, which can lead to heart attacks and strokes. Heart disease is becoming a global threat to the world due to people’s unhealthy lifestyles, prevalent stroke history, physical inactivity, and current medical background. In predictive analytics, many studies were proposed to get alerts about forthcoming heart disease based on various attributes. However, the performance metrics were good, but the model was trained with few features. This study aims to train the model with all the essential attributes for heart disease prediction on Framingham Heart Study (FHS) dataset. The dataset is pre-processed with IQR (Inter Quartile Range) outlier detection followed by data oversampling using Synthetic Minority Oversampling Technique (SMOTE). We proposed a stack generalization approach, wherein various machine learning classifiers, namely logistic regression, random forest, K-nearest neighbour, Naïve Bayes, support vector machine, XGBoost, and decision tree with optimized hyperparameter trained the model to offer the best learner for the prediction of Coronary Heart disease with improved performance. The proposed model is tested on the original imbalanced and SMOTE-balanced FHS dataset. It is observed that logistic regression on the original (imbalanced) FHS dataset provides 86.51% accuracy, while the support vector machine on the SMOTE (balanced) FHS dataset outperformed the other models with an accuracy of 93.07%. Also, the proposed approach of stacked generalization with cross-validation provided 97.2% accuracy on SMOTE (balanced) dataset, which is remarkable. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

Ladder-Type Resonant Switched-Capacitor Converter Achieving 99.1% Peak Efficiency and 4750 w/in3Power Density for High-Current Application

引用

IEEE Transactions on Power Electronics 2024年第7期39卷 8040-8053页

作者： You, Dong-Ying Chiu, Huang-Jen Kuo, Shih-Hao National Taiwan University of Science and Technology Department of Electronic and Computer Engineering Taipei10607 Taiwan

A nonisolated fixed-ratio resonant switched-capacitor converter (RSCC) with high peak efficiency and high power density is encouraged as intermediate bus converter. With higher input voltages, stacked topologies achieve better component stress, but the relative current stress will be multiplied, e.g., ladder and Cockcroft-Walton topologies are not well suited for high-power high step-down ratio applications. Accordingly, in this article, a ladder-type RSCC is proposed to meet high current demands by combining symmetrical auxiliary resonant tanks with two interleaved phases. In addition to having a high-resolution scalability that satisfies any integer conversion ratio, the voltage stress on the flying capacitors does not increase significantly as the conversion ratio increases, making their capacitance minimally susceptible to component tolerances. Meanwhile, all mosfets are always well clamped at 1× output low-voltage level, which means that low-voltage devices are permitted to be used, resulting in a good factor of merit. By combining adaptive constant on-time with dead-time modulation, an effective duty cycle is optimized, and controlled current sharing is achieved, making multistage parallel expansion very flexible and efficient. To demonstrate the feasibility and practicality, a compact 3-to-1 hardware prototype was realized with a load current of up to 75 A, resulting in an excellent power density of 4750 W/in3. A peak efficiency of 99.1% and a full-load efficiency of 98.1% were also achieved. © 1986-2012 IEEE.

关键词： Topology

来源：评论

学校读者我要写书评

暂无评论

SLOG: smartphone-based semantic lifelogging framework for digital prosthetic memory development

引用

Personal and Ubiquitous Computing 2023年第6期27卷 2071-2102页

作者： Ali, Shaukat Khusro, Shah Khan, Akif Khan, Inayat Alam, Iftikhar Solehria, Salman Faiz Department of Computer Science University of Peshawar Peshawar25120 Pakistan Department of Computer Science Sarhad University of Information Technology Peshawar25000 Pakistan Department of Computer Science University of Engineering and Technology Mardan Mardan23200 Pakistan

The advancements in sensors, processing, storage and networking technologies have turned smartphone into de facto lifelogging device. Realizing the lifelogging potential of smartphone, researchers have postulated with smartphone-based lifelogging systems with varying capabilities of capturing and storing information about peoples’ daily life activities, behaviors, interactions, locations, environments, etc., into their lifelog archives. However, the systems are lacking with several capturing, storing, retrieving and design issues, importantly not organizing the lifelog information with the same semantics as they exist in the real world and encoded in human episodic memory. The semantic organization of lifelog information will develop a true digital prosthetic memory to solve problems in different real-world domains including memory augmentation. In this paper, we have presented the architecture and implementation of Semantic LifelOGging (SLOG) framework for smartphone-based semantic lifelogging using Semantic Web technologies. The SLOG framework incorporates smartphone sensors and processing capabilities, and Semantic Web technologies to develop a semantically enriched digital prosthetic memory on smartphone. The semantic lifelog data model is developed by extending SmartOntoSensor ontology with relevant concepts and properties to represent, store and relate lifelog and contextual information. The semantic lifelog data model will provide contextual semantics to organize, annotate and relate lifelog information like human episodic memory and provide enough contextual cues to recall lifelog information like associative recall in human memory. The proof-of-concept applications are developed to show practicality of the SLOG framework. The quantitative and empirical evaluations have shown the SLOG framework and system has improved features and is acceptable in usability and effectiveness. © 2022, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part

关键词： Smartphones

来源：评论

学校读者我要写书评

暂无评论

A Channel Usage Probability Awareness Anti-Jamming Channel Hopping Protocol for Cognitive Radio Networks

引用

IEEE Transactions on Cognitive Communications and Networking 2024年第4期10卷 1327-1338页

作者： Lei, Sheng-Wei Chao, Chih-Min National Taiwan Ocean University Department of Computer Science and Engineering Keelung City202 Taiwan

To solve the rendezvous problem between any pair of Secondary Users (SUs) and potential jamming attacks in cognitive radio networks, many solutions have employed a channel hopping mechanism. The Primary User (PU) occupancy probability significantly affects the performance of these channel hopping protocols. Certain existing solutions adapt the frequency of channel usage based on predicted PU occupancy probability derived from historical averages. In this paper, we demonstrate the inadequacy of such predictions and propose a rendezvous-guaranteed anti-jamming channel hopping protocol with channel usage adjustment capability, termed OLAA_P. By using deep learning technology to predict future PU occupancy probability for each channel, OLAA_P can properly adjust the usage ratio of each channel. Simulation results validate that OLAA_P outperforms recently-proposed practical anti-jamming hopping protocols in terms of throughput. © 2015 IEEE.

关键词： Cognitive radio

来源：评论

学校读者我要写书评

暂无评论

A differential privacy-preserving federated learning scheme with predictive maintenance of wind turbines based on deep learning for feature compression and anomaly detection with state assessment

引用

Journal of Mechanical science and Technology 2024年第7期38卷 3413-3429页

作者： Chen, Huan Hsu, Hsin-Yao Hsieh, Jia-You Hung, Hsiang-En Department of Computer Science and Engineering National Chung Hsing University Taichung City40227 Taiwan

The transport and supply of traditional energy have been greatly affected, which has attracted the attention of countries worldwide and has become an urgent issue. Limited by its own traditional energy shortage and relying on many imported energy raw materials, Taiwan has gradually transformed natural resources into abundant renewable energy through energy transformation, of which wind power has increased up to 1.79 %. It mainly relies on the operation of wind turbines to convert wind power energy into electricity, so the operation of wind turbines is an important issue to maintain stable power collection. This study mainly discusses the wind turbine failure prediction model based on the supervisory control and monitoring system (SCADA) data of 31 wind turbines, and used deep learning and federated learning architecture combined with differential privacy method to find the most suitable model combination through experiments, and establishes wind turbines for state assessment of failure architecture with transmission and prediction, so as to improve wind turbine efficiency and reduce maintenance costs. The proposed method uses 1-dimensional convolutional neural network and bidirectional long short-term memory denoising autoencoder with the transformer (Conv1D-BiLSTM-DAE-TF) for feature compression, and then uses 1-dimensional convolutional neural network (Conv1D), recurrent neural network (RNN), long short term memory (LSTM) and gated recurrent unit (GRU) model for abnormal detection with state assessment. Combined with federated learning and differential privacy architecture for wind turbine failure prediction it can effectively predict performance and privacy security. The results show that it has the best for feature compression convergence performance and fault prediction based on differential privacy for protect model, and construction of models between different wind turbine equipment through the platform. © The Korean Society of Mechanical Engineers and Spring

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

High Performance MPI over the Slingshot Interconnect

引用

Journal of computer science & Technology 2023年第1期38卷 128-145页

作者： Kawthar Shafie Khorassani Chen-Chun Chen Bharath Ramesh Aamir Shafi Hari Subramoni Dhabaleswar K.Panda Department of Computer Science and Engineering The Ohio State UniversityColumbusOH 43210U.S.A.

The Slingshot interconnect designed by HPE/Cray is becoming more relevant in high-performance computing with its deployment on the upcoming exascale *** particular,it is the interconnect empowering the first exascale and highest-ranked supercomputer in the world,*** offers various features such as adaptive routing,congestion control,and isolated *** deployment of newer interconnects sparks interest related to performance,scalability,and any potential bottlenecks as they are critical elements contributing to the scalability across nodes on these *** this paper,we delve into the challenges the Slingshot interconnect poses with current state-of-the-art MPI(message passing interface)*** particular,we look at the scalability performance when using Slingshot across *** present a comprehensive evaluation using various MPI and communication libraries including Cray MPICH,Open-MPI+UCX,RCCL,and MVAPICH2 on CPUs and GPUs on the Spock system,an early access cluster deployed with Slingshot-10,AMD MI100 GPUs and AMD Epyc Rome CPUs to emulate the Frontier *** also evaluate preliminary CPU-based support of MPI libraries on the Slingshot-11 interconnect.

关键词： AMD GPU interconnect technology MPI(message passing interface) Slingshot

来源：评论

学校读者我要写书评

暂无评论

AnonMAKE: Toward Secure and Anonymous Mutually Authenticated Key Exchange Protocol for Vehicular Communications

引用

IEEE Transactions on Intelligent Transportation Systems 2024年第10期25卷 14245-14258页

作者： Karati, Arijit Chang, Li-Chun National Sun Yat-sen University Department of Computer Science and Engineering Kaohsiung80424 Taiwan

Due to real-time data sharing in the Intelligent Transportation System (ITS), the automobile industry has evolved enormously in recent years. However, as more vehicles become connected, the peril of cyberattacks increases, necessitating safeguarding traffic safety-related data provided by individual vehicles and public transportation to avoid perilous situations and infringement of confidentiality. In this regard, a recent work (10.1109/TITS.2021.3099488) suggested an anonymous batch data authentication and key exchange protocol. This paper shows that the work lacks true anonymity during authentication, employs a session-independent key, and is prone to session key disclosure and replay attacks. Apart from cryptanalysis, we present a novel privacy-enhancing mutual authentication and key agreement protocol supporting batch authentication resisting comprehensive security attacks in the vehicular system, including traffic data replay, vehicle impersonation, and man-in-the-middle attacks. The suggested scheme achieves security correctness under formal BAN logic analysis and is safer against the STRIDE effect. The empirical performance estimation reveals that the novel approach under adequate security assumptions outperforms the state-of-the-art solutions by achieving about 20% higher security traits while lowering 15% authentication delay, consuming 15% less bandwidth, and increasing 44% storage efficiency. Thus, our protocol is more suitable for lightweight vehicular communications. © 2000-2011 IEEE.

关键词： Vehicular ad hoc networks

来源：评论

学校读者我要写书评

暂无评论

Advancing emotion recognition via EEG signals using a deep learning approach with ensemble model

引用

Journal of Intelligent and Fuzzy Systems 2024年第1-2期47卷 143-154页

作者： Immanuel, Rajeswari Rajesh Sangeetha, S.K.B. Department of Computer Science and Engineering SRM Institute of Science and Technology Faculty of Engineering and Technology Vadapalani Campus Tamil Nadu Chennai India

Human emotions are the mind's responses to external stimuli, and due to their dynamic and unpredictable nature, research in this field has become increasingly important. There is a growing trend in utilizing deep learning and machine learning techniques for emotion recognition through EEG (electroencephalogram) signals. This paper presents an investigation based on a real-time dataset that comprises 15 subjects, consisting of 7 males and 8 females. The EEG signals of these subjects were recorded during exposure to video stimuli. The collected real-time data underwent preprocessing, followed by the extraction of features using various methods tailored for this purpose. The study includes an evaluation of model performance by comparing the accuracy and loss metrics between models applied to both raw and preprocessed data. The paper introduces the EEGEM (Electroencephalogram Ensemble Model), which represents an ensemble model combining LSTM (Long Short-Term Memory) and CNN (Convolutional Neural Network) to achieve the desired outcomes. The results demonstrate the effectiveness of the EEGEM model, achieving an impressive accuracy rate of 95.56%. This model has proven to surpass the performance of other established machine learning and deep learning techniques in the field of emotion recognition, making it a promising and superior tool for this application. © 2024 - IOS Press. All rights reserved.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Comparative Analysis of Diverse Architectures for Accurate Blood Cancer Cell Classification

Comparative Analysis of Diverse Architectures for Accurate B...

引用

2024 International Conference on computer, Electrical and Communication engineering, ICCECE 2024

作者： Md. Muttakin, G.K. Yesmin, Farhana Alam, Syeda Sadia Ashraf, Md Sadi Akuthota, Vishwanath Quader, Montaser Abdul Rajshahi University of Engineering and Technology Department of Electrical and Computer Engineering India Metropolitan University Department of Computer Science and Engineering United Kingdom Brac University Department of Computer Science and Engineering Bangladesh Drpinnacle Hyderabad India Green University of Bangladesh Department of Computer Science and Engineering Bangladesh

ISBN: (纸本)9798350386479

Blood cancer cell diagnosis is crucial in medical diagnostics. It demands accurate classification of blood cell images. Proper classification of blood cancer cells is fundamental for accurately diagnosing the specific type and subtype of blood cancer as well as efficient treatment planning. It provides specific information, helping healthcare professionals predict the prognosis, survival rates, and the potential for disease recurrence. On this basis, deep learning models have demonstrated remarkable performance. In this paper, we have introduced a comprehensive research into blood cancer cell classification, employing a diverse dataset encompassing various blood cell types. We explore the effectiveness of VGG19 with batch normalization, Vision Transformer (ViT), Ensemble Adversarial Inception-ResNetV2, DenseNet201, and ResNeXt50 architectures in this challenging task. In order to enhance the model performance, we integrate essential data preprocessing techniques, including resizing, cropping, and normalization. Additionally, novel data augmentation strategies, such as random cropping, and flipping are introduced to augment the training dataset and improve model generalization. Remarkably, VGG19 with batch normalization has shown tremendous success by achieving 99.86% accuracy. Moreover, the DenseNet201 model has performed brilliantly by achieving an accuracy of 98.55%. The ResNeXt50 model shows excellent performance with an accuracy of 98.31%. The Vision Transformer (ViT) achieves a solid accuracy of 98.91%. Lastly, Ensemble Adversarial Inception-ResNet V2 also performed no-tably achieving 96.38%. In this context, VGG19 with batch normalization is able to show more excellent performance than other models by achieving the highest accuracy. © 2024 IEEE.

关键词： Cancer cells

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：