检索结果-内蒙古大学图书馆

Optimized featured swarm convolutional neural network (OFSCNN) model based dialect recognition system for Bagri Rajasthani language

引用

International Journal of Information Technology (Singapore) 2025年第1期17卷 121-133页

作者： Kukana, Poonam Sharma, Pooja Bhardwaj, Neeru Department of Computer Science and Engineering University School of Engineering & amp Technology Rayat-Bahra University Punjab Mohali India Department of Computer Science and Engineering Chandigarh University Punjab Mohali India

The dialects of a language hold a significant place in speech processing (SP) applications. The objective of dialect identification is to categorize speech sample data into a specific dialect of a speaker's spoken language. A dialect recognition system must effectively distinguish between different dialects of a standard language, as they tend to possess many similarities. The dialect of a language is not a distinct characteristic, as it can be impacted by the utterer's birthplace. Researchers in the domain of automatic speech recognition (ASR) face difficulties in identifying the speech patterns unique to each dialect or language. The proposed work recognizes the dialects of the Bagri Rajasthani language from undefined expressions of speech. Rajasthani language is one of the eldest and most famous languages in the Indo-Aryan languages. It comprises the different dialects and for recognizing the dialects, it used dissimilar phases of acoustic and spectral characteristics of the speech signal (SS). The spectral and acoustic features of SSs are measured to design the system. As there is no specific speech dataset for Bagri dialects, the database is built, to verify the Bagri dialects of the Rajasthani language. To improve the accuracy rate, and error rate in recognizing the Bagri dialects, the acoustic and spectral characteristics of speech expressions are joined. To verify several Bagri dialects of the Rajasthani language, different simulations for classification and investigation are carried out i.e., OFSCNN model, GA-NN, etc. The outcomes are important and the accuracy of 96.95% for the OFSCNN model, 80.63% for GA-NN, and 93.45% for the SVM method is an achievement. © Bharati Vidyapeeth's Institute of computer Applications and Management 2024.

关键词： CNN (convolutional neural network Dialects MFCC (Mel frequency co-efficient cepstral) OFSCNN (optimized featured swarm convolutional neural network) PSO (particle swarm optimization) Speech Dialect Recognition

来源：评论

学校读者我要写书评

暂无评论

Network Traffic Matrix Estimation Based on Link Loads Sampling

引用

IEEE Transactions on Network Science and engineering 2025年第3期12卷 1524-1539页

作者： Sun, Wenyue Chen, Qian Song, Xuehua Bertino, Elisa Wang, Changda Jiangsu University School of Computer Science and Communication Engineering Zhenjiang212013 China Yangzhou Polytechnic College School of Information Engineering Yangzhou225009 China Purdue University Department of Computer Science West LafayetteIN47907 United States

Traffic Matrix (TM) represents traffic between all Origin-Destination (OD) node pairs in a network, playing a crucial role in network management. While the methods for TM acquisition typically require scaling each link load in a network, measurement costs arise drastically as the number of links grows exponentially with the number of nodes. To address this issue while achieving TM with acceptable accuracy aligned with various network management requirements, the paper proposes two novel methods. The first is the ME-TMEIM (TM Estimation with Incomplete Measurement Based on Maximum Entropy) method, an efficient approach with yet acceptable accuracy. The second is the D-TMEIM (Dynamic TMEIM) method. Compared to the ME-TMEIM method, the D-TMEIM method trades the TM's acquisition accuracy for efficiency. It adds temporal constraints on network traffic to improve the precision of the obtained missing link loads, based on which the TM is generated using the CS-OMP (Compressed Sensing-Orthogonal Matching Pursuit) algorithm. Experimental results using publicly available Abilene and GÉANT networks demonstrate that the proposed methods not only enhance TM acquisition efficiency but also maintain nearly the same accuracy as the known methods that acquire TM through exhaustive links measurement. © 2013 IEEE.

关键词： Efficiency

来源：评论

学校读者我要写书评

暂无评论

IoT-Aware Real-Time Healthcare Diagnostic Framework For Diabetes Using Wearable Sensors through Deep Reinforcement Learning

引用

IEEE Internet of Things Journal 2025年第11期12卷 18183-18194页

作者： Farman, Haleem Shahzad, Yasir Jan, Bilal Nasralla, Moustafa M. Sallam, Karam M. Munasinghe, Kumudu Jamalipour, Abbas Prince Sultan University Smart Systems Engineering Lab Department of Communications and Networks Engineering College of Engineering Riyadh Saudi Arabia University of Peshawar Department of Computer Science Pakistan FATA University Department of Computer Science Kohat Pakistan University of Sharjah Computer Science Department Sharjah United Arab Emirates University of Canberra School of Information Technology and Systems ACT2617 Australia The University of Sydney School of Electrical and Information Engineering NSW2006 Australia

Machine Learning (ML) with 5G technology has revolutionized smart healthcare. It has helped improve the quality of care, such as real-time analysis, decision-making, patient monitoring, and personalized treatments. In this paper, a 5G aware real-time diabetes prediction framework is proposed using optimized Bidirectional Long Short-Term Memory (Bi-LSTM) with Deep Reinforcement Learning (DRL). Bi-LSTM can analyze time-series data in forward and backward passes on patient health metrics, such as Blood Glucose (BG) levels of a diabetic patient, to identify patterns and trends and make predictions about future health outcomes. A local dataset using a wearable sensor is collected of 10 Type-2 diabetic patients, encompassing daily BG levels at different times, alongside additional parameters such as blood pressure and body weight. The proposed framework leverages a dataset of 1,830 data points to forecast glucose levels for the following day. It also harnesses DRL to improve and optimize the model's future predictive performance. The obtained results are evaluated with other ML algorithms to validate the effectiveness of the proposed framework, which shows an improvement from 93.1% to 98.6% accuracy. The patient's diabetic condition is categorized using the Surveillance Error Grid (SEG) to increase the clinical impact of glucose prediction and make informed decisions. The results show that Bi-LSTM-DRL is an effective approach to predicting glucose levels in real-time and can adapt to health-related changes and optimize its predictions accordingly. © 2014 IEEE.

关键词： Deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

ViT-SENet-Tom: machine learning-based novel hybrid squeeze–excitation network and vision transformer framework for tomato fruits classification

引用

Neural Computing and Applications 2025年第9期37卷 6583-6600页

作者： Swapno, S M Masfequier Rahman Nobel, S. M. Nuruzzaman Islam, Md Babul Bhattacharya, Pronaya Mattar, Ebrahim A. Department of CSE Bangladesh University of Business and Technology Dhaka1216 Bangladesh Department of Computer Modeling Electronic and System Engineering UNICAL Rende Italy Department of Computer Science and Engineering Amity School of Engineering and Technology and Research and Innovation Cell Amity University West Bengal Kolkata India Robotics - Cybernetic College of Engineering University of Bahrain Manama Bahrain

Tomatoes are essential fruits in numerous nations for their vast demand. It is very important to maintain the freshness of tomatoes. One of the primary challenges in the recent culinary landscape is accurately identifying healthy tomatoes while effectively eliminating damaged or rejected ones. Existing approaches employ various strategies for categorizing tomato fruit, but they often suffer from inaccuracies, slow detection, and suboptimal performance. Thus, motivated by this gap, in this paper, we propose a novel machine learning (ML) framework, ViT-SENet-Tom, which is a hybrid vision transformer (ViT) model with squeeze and excitation (SENet) block network for fast, accurate, and efficient tomato fruit classification. The framework works on three tomato classes, respectively, the ripe, unripe, and reject. In developing the proposed model, we utilized advanced and newly designed layers and functions. This integration created a more complex and sophisticated neural network, significantly enhancing efficiency and contributing to the model’s novelty. Our chosen dataset was small initially, but we implemented augmentation techniques to increase its size. This approach made our system more reliable, efficient, and effective. The hybrid ViT-SENet framework employs encoders and self-attention networks with squeeze and excitation channel functions to allow precise, robust, fast, and efficient tomato classification. In simulation, the framework achieves a training accuracy of 99.87% and validation accuracy of 93.87%, indicating the precise classification of tomatoes. Besides, this work tests accuracy using fivefold cross-validation. The highest accuracy seen at fold-5 is 99.90%. These testing results demonstrate the efficacy of the proposed framework in real-deployment scenarios. The implementation has the potential to provide enhanced and more sustainable food security and safety in future. © The Author(s) 2025.

关键词： Fruits

来源：评论

学校读者我要写书评

暂无评论

Intelligent Guitar Chord Recognition Using Spectrogram-Based Feature Extraction and AlexNet Architecture for Categorization

引用

International Journal of Advanced computer Science and Applications 2025年第4期16卷 758-767页

作者： Korade, Nilesh B. Salunke, Mahendra B. Bhosle, Amol A. Sangve, Sunil M. Joshi, Dhanashri M. Asalkar, Gayatri G. Kadu, Sujata R. Sarwade, Jayesh M. Department of Computer Engineering JSPM’s Rajarshi Shahu College of Engineering Tathawade Maharashtra Pune411033 India Department of Computer Engineering PCET’s Pimpri Chinchwad College of Engineering and Research Ravet Maharashtra Pune412101 India Department of Computer Science and Engineering School of Computing MIT Art Design and Technology University Loni Kalbhor Maharashtra Pune412201 India Department of Artificial Intelligence and Data Science Vishwakarma Institute of Technology Bibwewadi Pune411037 India Department of Computer Engineering Terna Engineering College Maharashtra Mumbai India Department of Information Technology JSPM’s Rajarshi Shahu College of Engineering Tathawade Maharashtra Pune411033 India

Chord prediction plays a key role in the advancement of musical technological innovations, such as automatic music transcription, real-time music tutoring, and intelligent composition tools. Accurate chord prediction can assist musicians, educators, and developers in constructing tools that help in learning, playing, and composing music. Background noise and audio distortions may have an impact on chord prediction accuracy, particularly in real-world situations. Chords can have distinct voicings or finger positions on the guitar, resulting in slight variations in audio representation. This study focuses on the classification of guitar chords using techniques of deep learning. There are eight major and minor guitar chords in the dataset. They have been turned into spectrograms, chromagrams, and Mel Frequency Cepstral Coefficients (MFCC) so that features can be extracted. Various deep learning architectures, including CNN, ResNet50, AlexNet, and VGG, were employed to classify the chords. Experimental results demonstrated that the spectrogram-based AlexNet model outperforms others, achieving good accuracy and robustness in chord classification. The proposed study demonstrates the efficiency of spectrograms and advanced deep learning models for audio signal processing in music applications. By automating chord detection, this study provides beneficial resources for music learners as well as educators, enabling more efficient learning and real-time feedback during practice sessions. © (2025), (Science and Information Organization). All Rights Reserved.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Fuzzy relevance vector machine-based segmentation and adaptive network-based fuzzy inference systems for Parkinson’s disease detection and classification

引用

Discover Computing 2025年第1期28卷 1-18页

作者： Jayasudha, V. Deepa, N. T, Devi Department of Computer Science and Engineering Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Chennai India

Parkinson's disease (PD) is the second most prevalent neurodegenerative condition, yet the use of EEG signals for its differentiation from healthy controls remains underexplored. This paper presents a novel approach that leverages complex, non-stationary, and non-linear EEG signals for PD detection. We employ Butterworth filtering for preprocessing, followed by fuzzy relevance vector machine segmentation to effectively manage the extensive data points. Key features are extracted using a fast scale invariant feature transform, and classification is performed using an adaptive network-based fuzzy inference system (ANFIS). Our proposed method achieves an exceptional classification accuracy of 99.2%, significantly outperforming existing techniques. This work not only enhances the diagnostic accuracy for early-stage PD but also addresses the challenges of data management in clinical settings, thereby contributing to the development of more efficient diagnostic tools in neurology. © The Author(s) 2025.

关键词： Adaptive network based fuzzy interference system Butterworth filtering EEG signals Fast scale invariant feature transform Fuzzy relevance vector machine segmentation Parkinson disease

来源：评论

学校读者我要写书评

暂无评论

Privacy-preserving explainable AI: a survey

引用

Science China(Information Sciences) 2025年第1期68卷 23-56页

作者： Thanh Tam NGUYEN Thanh Trung HUYNH Zhao REN Thanh Toan NGUYEN Phi Le NGUYEN Hongzhi YIN Quoc Viet Hung NGUYEN School of Information and Communication Technology Griffith University School of Computer and Communication Sciences Ecole Polytechnique Federale de Lausanne Faculty of Mathematics and Computer Science University of Bremen Faculty of Information Technology HUTECH University Department of Computer Science Hanoi University of Science and Technology School of Electrical Engineering and Computer Science The University of Queensland

As the adoption of explainable AI(XAI) continues to expand, the urgency to address its privacy implications intensifies. Despite a growing corpus of research in AI privacy and explainability, there is little attention on privacy-preserving model explanations. This article presents the first thorough survey about privacy attacks on model explanations and their countermeasures. Our contribution to this field comprises a thorough analysis of research papers with a connected taxonomy that facilitates the categorization of privacy attacks and countermeasures based on the targeted explanations. This work also includes an initial investigation into the causes of privacy leaks. Finally, we discuss unresolved issues and prospective research directions uncovered in our analysis. This survey aims to be a valuable resource for the research community and offers clear insights for those new to this domain. To support ongoing research, we have established an online resource repository, which will be continuously updated with new and relevant findings.

关键词： privacy-preserving explainable AI privacy attacks privacy defences PrivEx PPXAI

来源：评论

学校读者我要写书评

暂无评论

Electrochemical sensor for dopamine detection based on multiwalled carbon nanotube/molybdenum disulfide quantum dots with machine learning integration and anti-interference capability

引用

Tungsten 2025年第1期7卷 137-151页

作者： Geethukrishnan Onkar Apte Kiran Kumar Tadi Energy Electrodics and Electrocatalysis (EEE) Lab Department of Chemistry School of Advanced Sciences Vellore Institute of Technology School of Computer Science and Engineering Vellore Institute of Technology

The precise detection and measurement of dopamine（DA）,a crucial neurotransmitter in the human body,plays a significant role in diagnosing,preventing,and treating neurological diseases associated with its levels.A highly sensitive DA electrochemical sensor was constructed by combining molybdenum disulfide quantum dots（MSQDs） with multiwalled carbon nanotubes（MWCNTs）.The MSQDs were synthesized using the shear exfoliation *** sensors consist of MSQDs with Mo-S edge catalytic centers for the DA redox reaction,and MWCNTs amplify the sensor *** linearity of the sensor for the detection of DA was tested in the presence of ascorbic acid(AA,50 μmol·L-1) and uric acid(UA,200 μmol·L-1),and exhibited linearity from 2 to 966 μmol·L-1of DA with 0.097 μA(mol·L-1)-1sensitivity and a low limit of detection of0.6 μmol·L-1（the ratio between signal and noise,S/N=3）.Moreover,the sensitivity and selectivity of the sensor were also studied using *** is no increase in amperometric current after adding the most potentially interfering *** sensor was successfully applied to recover DA in human blood sera ***,machine learning algorithms were operated to aid in the near-precise detection of DA in the heterogeneous mixture containing AA and *** algorithms facilitate the identification and quantification of DA amidst coexisting interferents,including AA,that are commonly present in biological matrices.

关键词： Dopamine MoS_2 Biosensor Multiwalled carbon nanotubes Machine learning Complex mixture

来源：评论

学校读者我要写书评

暂无评论

Comparison of Classical and Deep Learning-Based Feature Representations for Age-Related Macular Degeneration 5th

Comparison of Classical and Deep Learning-Based Feature Re...

引用

5th International Conference on Medical Imaging and computer Aided Diagnosis, MICAD 2024

作者： Sinichi, Parsa Bernabeu, Miguel O. Javidi, Malihe Department of Computer Engineering Faculty of Electronic and Computer Engineering Quchan University of Technology Quchan Iran Centre for Medical Informatics University of Edinburgh Edinburgh United Kingdom

ISBN: (纸本)9789819638628

Age-related Macular Degeneration (AMD) is a leading cause of visual impairment among the elderly worldwide. This study compares deep learning-based and classical feature extraction methods for AMD classification using colour fundus images, a cost-effective imaging modality. This comparative approach aims to evaluate the performance of advanced deep learning methods against traditional computer vision approaches in generating effective feature representation. To achieve this, we compared various classical feature extraction methods alongside deep learning-based approaches, specifically VGG16 pre-trained on ImageNet, RETFound and RETFound-Green, which are foundation models pre-trained on a significant number of retinal images. The results demonstrate that combinations of classical features still obtained notable performance and even outperformed the VGG16 model pre-trained on ImageNet. These results also highlight the need for deep learning models to be trained on domain-specific data to surpass classical approaches. In this regard, the pre-trained RETFound and RETFound-Green models yielded promising outcomes for AMD classification, and the highest performance was achieved using a fine-tuned RETFound model. Furthermore, the best proposed method was compared with the state-of-the-art methods. This approach achieved competitive results on the ADAM dataset, demonstrating its robustness and effectiveness in AMD classification while utilising a significantly simpler architecture. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Resilient generation planning considering long periods of low-RES output

引用

Environment Systems and Decisions 2025年第2期45卷 1-13页

作者： Stasinos, Ektor-Ioannis Panteli, Mathaios Hatziargyriou, Nikos School of Electrical and Computer Engineering National Technical University of Athens Athens Greece Department of Electrical and Computer Engineering University of Cyprus Nicosia Cyprus

The growing concerns over mitigating climate change effects resulted in power system planning and generation expansion strategies that aim in increasing penetration of intermittent renewable energy sources (RES) to fulfill the national energy and climate plans (NECPs) of countries worldwide. However, the variable and intermittent nature of RES output, compounded by uncertainties arising from climatic and weather changes, poses significant challenges. These challenges necessitate careful consideration in generation planning and in determining the operational reserves of RES-rich power systems to ensure resilience. This paper employs a generation capacity operational planning method based on the Screening Curve Method (SCM), assuming various levels of large RES penetration and diverse spinning generation reserve policies, necessary to cope with various levels of RES generation deficits, to guarantee power system resilience. The proposed approach can be used to assess the overall reduction of capacity and decommissioning of thermal units, while ensuring resilient operation against various levels of annual RES generation deficits. Furthermore, an approximation of the total annual system operation cost is also provided, considering the impact of the employed spinning reserve policies. © The Author(s) 2025.

关键词： Decommissioning of thermal units Low-renewable energy sources generation Power system resilient generation planning Screening curve method

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：