检索结果-内蒙古大学图书馆

A new multi-stream approach using acoustic and visual features for robust speech recognition system

Materials Today: Proceedings 2022年 62卷 4916-4924页

作者： Radha, N. Shahina, A. Nayeemulla Khan, A. Velusami, Jansi Rani Sella Department of Information Technology Sri Sivasubramaniya Nadar College of Engineering Chennai India School of Computing Science and Engineering Vellore Institute of Technology Chennai India Department of Computing Science and Engineering Sri Sivasubramaniya Nadar College of Engineering Chennai India

Building a robust Automatic Speech Recognition (ASR) system and improving recognition accuracy in adverse conditions is still a challenging task. One way to improve the robustness of an ASR system is combining information from multiple sources (streams). A multi-stream approach which handles the multiple inputs at the model level is the key contribution of our work. Standard mic (Sm), Throat mic (Tm), and Lip reading (Lr) are the various source streams that have been used. This work explores a static weighted two stream HMM (TSH) and multi-stream HMM (MSH) model for the bimodal and multimodal systems. A syllabic units of the Hindi language database categorized into three categories – Vowel, Place of Articulation (POA), and Manner of Articulation (MOA) are used for training and testing. In this study, four types of TSH are proposed for the combination of bimodal ((Sm+Tm), (Tm+Lr), (Sm+Lr), (Lm+Lm)) systems and one type MSH is proposed for multimodal (Sm+Tm+Lr) system in a synchronous and asynchronous manner. Mel Frequency Cepstral Coefficient (MFCC) features are used for Sm and Tm signals. Combined pixel-motion based features (DCT/DWT-MHI) are used for Lr signals. Among these two features, DWT outperforms than DCT and used as a feature for visual speech. Experiments were conducted for bimodal and multimodal system. The proposed MSH approach shows improvements of 1.36%, 6.21%, and 5.8% in recognition accuracy for Vowel, POA and MOA category respectively, as compared to bimodal systems. © 2022

关键词： Discrete wavelet transforms

来源：评论

学校读者我要写书评

暂无评论

Few-Shot Learning for Early Detection of Laryngeal Cancer with Explainable AI 8

Few-Shot Learning for Early Detection of Laryngeal Cancer wi...

引用

8th International Conference on Electronics, Materials engineering and Nano-technology, IEMENTech 2025

作者： Pradhan, Ujan Prerna Aditya, M.N. SRM Institute of Science & Technology Department of Computing Technologies Chennai India SRM Institute of Science & Technology Department of Biomedical Engineering Chennai India SRM Institute of Science & Technology Department of Electronics and Communication Engineering Chennai India

ISBN: (纸本)9798331520762

Early diagnosis of laryngeal cancer significantly relates to prognosis and treatment. This work presents a classification model that uses very few-shot learning where classification between healthy and early-stage cancerous tissues can be done accurately even with very limited data. The proposed approach relies on an expert-curated dataset of 1,320 image patches of size 100x100 pixels, which suggests promising potential for real-world applications, especially in strictly limited amounts of annotated medical images. This approach uses Explainable AI (XAI) to increase the transparency of the model and eventually enhance clinical trust in the predictions obtained with that model. This technique has the promise for early detection, but it also has high accuracy and further allows clinical thinking into the decision mechanism of the model. That is a future direction for developing the refined version of such a technique to develop robustness and application in different clinical settings. © 2025 IEEE.

关键词： Mammography

来源：评论

学校读者我要写书评

暂无评论

A Recurrent Neural Network for Multimodal Anomaly Detection by Using Spatio-Temporal Audio-Visual Data

引用

Computers, Materials & Continua 2024年第11期81卷 2493-2515页

作者： Sameema Tariq Ata-Ur-Rehman Maria Abubakar Waseem Iqbal Hatoon S.Alsagri Yousef A.Alduraywish Haya Abdullah AAlhakbani Department of Electrical Engineering University of Engineering and TechnologyLahore54890Pakistan Department of Electrical Engineering National University of Science and TechnologyNational University of Sciences and TechnologyIslamabad24090Pakistan Department of Business and Computing Ravensbourne University LondonRavensbourne UniversityLondonSE100EWEngland Electrical and Computer Engineering Department College of EngineeringSultan Qaboos UniversityMuscat123Oman College of Computer and Information Sciences Imam Mohammad Ibn Saud Islamic University(IMSIU)Riyadh11673Saudi Arabia

In video surveillance,anomaly detection requires training machine learning models on spatio-temporal video ***,sometimes the video-only data is not sufficient to accurately detect all the abnormal ***,we propose a novel audio-visual spatiotemporal autoencoder specifically designed to detect anomalies for video surveillance by utilizing audio data along with video *** paper presents a competitive approach to a multi-modal recurrent neural network for anomaly detection that combines separate spatial and temporal autoencoders to leverage both spatial and temporal features in audio-visual *** proposed model is trained to produce low reconstruction error for normal data and high error for abnormal data,effectively distinguishing between the two and assigning an anomaly *** is conducted on normal datasets,while testing is performed on both normal and anomalous *** anomaly scores from the models are combined using a late fusion technique,and a deep dense layer model is trained to produce decisive scores indicating whether a sequence is normal or *** model’s performance is evaluated on the University of California,San Diego Pedestrian 2(UCSD PED 2),University of Minnesota(UMN),and Tampere University of technology(TUT)Rare Sound Events datasets using six evaluation *** is compared with state-of-the-art methods depicting a high Area Under Curve(AUC)and a low Equal Error Rate(EER),achieving an(AUC)of 93.1 and an(EER)of 8.1 for the(UCSD)dataset,and an(AUC)of 94.9 and an(EER)of 5.9 for the UMN *** evaluations demonstrate that the joint results from the combined audio-visual model outperform those from separate models,highlighting the competitive advantage of the proposed multi-modal approach.

关键词： Acoustic-visual anomaly detection sequence-to-sequence autoencoder reconstruction error late fusion regularity score

来源：评论

学校读者我要写书评

暂无评论

Feature Selection with Deep Reinforcement Learning for Intrusion Detection System

引用

Computer Systems Science & engineering 2023年第9期46卷 3339-3353页

作者： S.Priya K.Pradeep Mohan Kumar Department of Computer Science and Engineering SRM Institute of Science and TechnologyRamapuramChennai600089India Department of Computing Technologies SRM Institute of Science and TechnologyKattankulathurChennai603203India

An intrusion detection system(IDS)becomes an important tool for ensuring security in the *** recent times,machine learning(ML)and deep learning(DL)models can be applied for the identification of intrusions over the network *** resolve the security issues,this paper presents a new Binary Butterfly Optimization algorithm based on Feature Selection with DRL technique,called BBOFS-DRL for intrusion *** proposed BBOFSDRL model mainly accomplishes the recognition of intrusions in the *** attain this,the BBOFS-DRL model initially designs the BBOFS algorithm based on the traditional butterfly optimization algorithm(BOA)to elect feature ***,DRL model is employed for the proper identification and classification of intrusions that exist in the ***,beetle antenna search(BAS)technique is applied to tune the DRL parameters for enhanced intrusion detection *** ensuring the superior intrusion detection outcomes of the BBOFS-DRL model,a wide-ranging experimental analysis is performed against benchmark *** simulation results reported the supremacy of the BBOFS-DRL model over its recent state of art approaches.

关键词： Intrusion detection security reinforcement learning machine learning feature selection beetle antenna search

来源：评论

学校读者我要写书评

暂无评论

Multi-lingual news stack based ensemble classification for deep news feed analysis

引用

Neural computing and Applications 2025年 1-12页

作者： Sakthivel, Rakesh Kumar Nagasubramanian, Gayathri Sankayya, Muthuramalingam Kadry, Seifedine Rocha, Alvaro Karnataka Bengaluru India Department of Information Technology Thiagarajar College of Engineering Madurai India Faculty of Applied Computing and Technology Noroff University College Kristiansand Norway Alvaro Rocha ISEG University of Lisbon Lisbon Portugal

Analyzing and extracting effective words of news feeds are critical tasks. News making sources may feed their news articles in different manners based on their own perspective. In this situation, finding likelihood samples from the vast news feeds is an essential task. Moreover, acquiring optimal word deviations, news varieties, sentiment metrics, and multi-lingual analysis are major tasks of this proposed system. To achieve this massive news feed analysis task, the proposed system uses specially designed multi-classifier units and news stack manipulation procedures. The proposed system integrates different classifiers and dynamic news stacks to handle datasets of multi-lingual attributes. The results show that the proposed method performs with better accuracy rate than existing techniques. This gives 10 to 15% higher accuracy rate than existing works. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Data accuracy

来源：评论

学校读者我要写书评

暂无评论

AFuNet: an attention-based fusion network to classify texts in a resource-constrained language

引用

Neural computing and Applications 2025年第9期37卷 6725-6748页

作者： Hossain, Md. Rajib Hoque, Mohammed Moshiul Dewan, M. Ali Akber Hoque, Enamul Siddique, Nazmul Department of Computer Science & Engineering Chittagong University of Engineering & Technology Chittagong4349 Bangladesh School of Computing and Information Systems Faculty of Science and Technology Athabasca University AthabascaABT9S 3A3 Canada School of Information Technology York University TorontoONM3J 1P3 Canada School of Computing Engineering and Intelligent Systems Ulster University Coleraine United Kingdom

In the era of widespread Internet use and extensive social media interaction, the digital realm is accumulating vast amounts of unstructured text data. This unstructured data often contain undesirable information, necessitating time-consuming manual classification efforts. An intelligent text classification system capable of automatically categorizing digitized texts based on semantic meaning is crucial. However, this task is particularly challenging for low-resource languages like Bengali due to a shortage of annotated corpora, issues with out-of-vocabulary words, lack of domain-specific hyperparameter tuning, limited ability to extract generalized text features, and class imbalances within the corpus. AFuNet: an attention-based fusion network to classify texts in a resource-constrained language. AFuNet undergoes a comprehensive four-phase experimental process, including baseline model evaluation and hyperparameter tuning, late fusion and model selection, attention-based early fusion and model identification, and an ablation study with impact analysis. Fine-tuned based on five Bengali text classification corpora, AFuNet achieves impressive accuracies: 96.60 ± 0.2 (BTCC11), 85.37 ± 0.2 (OSBC), 97.35 ± 0.2 (BARD), 93.74 ± 0.2 (IndicNLP), and 96.51 ± 0.2 (ProthomAlo). In comparison with previous state-of-the-art models on these corpora, AFuNet demonstrates significant accuracy improvements ranging from 0.54% to 4.49%, showcasing its effectiveness in advancing text classification capabilities for the Bengali language. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Determination and Classification of Diabetes Mellitus Applying Ensemble Machine Learning Classifiers 1

Determination and Classification of Diabetes Mellitus Applyi...

引用

1st International Conference on Cognitive, Green and Ubiquitous computing, IC-CGU 2024

作者： Patel, Siddharth Hassan, Zayd Iniyan, S. SRM Institute of Science and Technology Department of Computer Science and Engineering Kaatankulathur Tamilnadu Chennai India School of Computing SRM Institute of Science and Technology Department of Computing Technologies Kaatankulathur Tamilnadu Chennai India

ISBN: (纸本)9798350361780

This paper addresses the accurate and early prediction of the onset and progression of diabetes, an important aspect in the management of this chronic metabolic disease given a variety of information like the history of the diseased person and his day-to-day routine factors by using state-of-the-art machine learning techniques to obtain predictive dynamic models. The proposed work not only demonstrates the effectiveness of different algorithms but also highlights the importance of feature engineering to refine predictions. This work uses ensemble classifiers and voting classifier algorithms to develop better accuracy for diabetes prediction. Combining the results of multiple base classifiers like KNN, SVM, RFC, and decision tree ensemble methods aims to enhance the predictive ability and chances of improvement in the patient. The treatment adopted is different from patient to patient based on the result of the treatment. In this research, we utilize a combination of ensemble techniques and a voting classifier to construct a comprehensive diabetes prediction model. This model is evaluated using real-world data. Our findings reveal the exceptional effectiveness of ensemble classifiers and the voting classifier, underscoring their potential to achieve precision and the possibility of recovery and probability. These outcomes not only enable early intervention but also offer valuable insights into the care of individual patients. © 2024 IEEE.

关键词： Decision trees

来源：评论

学校读者我要写书评

暂无评论

Advancing Air Quality Monitoring: TinyML-Based Real-Time Ozone Prediction with Cost-Effective Edge Devices 2nd

Advancing Air Quality Monitoring: TinyML-Based Real-Time Ozo...

引用

2nd International Conference on Intelligent Manufacturing and Robotics, ICIMR 2024

作者： Ken, Huam Ming Behjati, Mehran Rafsanjani, Ahmad Sahban Aslam, Saad Meng, Yap Kian Majeed, Anwar P. P. Abdul Zheng, Yufan Department of Computing and Information Systems School of Engineering and Technology Sunway University Selangor Bandar Sunway47500 Malaysia Department of Computing and Information Systems School of Engineering and Technology Sunway University Selangor Bandar Sunway47500 Malaysia Suzhou Taicang215400 China

ISBN: (纸本)9789819639489

The escalation of urban air pollution necessitates innovative solutions for real-time air quality monitoring and prediction. This paper introduces a novel TinyML-based system designed to predict ozone concentration in real-time. The system employs an Arduino Nano 33 BLE Sense microcontroller equipped with an MQ7 sensor for carbon monoxide (CO) detection and built-in sensors for temperature and pressure measurements. The data, sourced from a Kaggle dataset on air quality parameters from India, underwent thorough cleaning and preprocessing. Model training and evaluation were performed using Edge Impulse, considering various combinations of input parameters (CO, temperature, and pressure). The optimal model, incorporating all three variables, achieved a mean squared error (MSE) of 0.03 and an R-squared value of 0.95, indicating high predictive accuracy. The regression model was deployed on the microcontroller via the Arduino IDE, showcasing robust real-time performance. Sensitivity analysis identified CO levels as the most critical predictor of ozone concentration, followed by pressure and temperature. The system’s low-cost and low-power design makes it suitable for widespread implementation, particularly in resource-constrained settings. This TinyML approach provides precise real-time predictions of ozone levels, enabling prompt responses to pollution events and enhancing public health protection. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Air quality

来源：评论

学校读者我要写书评

暂无评论

SMSecure: Leveraging Machine Learning for Smishing Detection 1st

SMSecure: Leveraging Machine Learning for Smishing Detection

引用

1st International Conference on Data engineering and Machine Intelligence, ICDEMI 2023

作者： Samad, Saleem Raja Abdul Balasubramaniyan, Sundaravadivazhagan Ganesan, Pradeepa Al-Kaabi, Amna Salim Ammaippan, Hariraman Sam, Jeyakumar Manickam Information Technology Department College of Computing and Information Sciences University of Technology and Applied Sciences-Shinas Shinas Oman Information Technology Department College of Computing and Information Sciences University of Technology and Applied Sciences-Al-Musannah Al-Musannah Oman Electrical Engineering Department College of Engineering and Technology University of Technology and Applied Sciences-Shinas Shinas Oman

ISBN: (纸本)9789819776153

The SMS facilitates the transmission of concise text messages between mobile phone users, serving a range of functions in both personal and business domains such as appointment confirmation, authentication, alerts, notifications, and banking updates. It plays a vital role in daily communication due to its accessibility, reliability, and compatibility. Its accessibility, reliability, and compatibility make it essential for daily communication. However, SMS unintentionally generates an environment where smishing can occur. This is because SMS is extensively available and reliable. Smishing attackers take advantage of this trust to deceive victims into divulging sensitive information or performing malicious actions. Early detection saves users from being victimized. Researchers introduced different methods for accurately detecting smishing attacks. Machine Learning models and Techniques for Language Processing are a promising approach for combating the escalating menace of SMS phishing attacks. By analyzing large datasets of SMS messages, machine learning models with natural language processing methods can differentiate between legitimate and fraudulent messages. To detect smishing attacks, this paper presents a method (SMSecure) that leverages machine learning models and language processing techniques. The results show that random forest and extreme gradient boosting attain higher accuracy levels. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Phishing

来源：评论

学校读者我要写书评

暂无评论

Adversarial Attacks Using Carlini Wagner

Adversarial Attacks Using Carlini Wagner

引用

2024 IEEE International Conference on Information technology, Electronics and Intelligent Communication Systems, ICITEICS 2024

作者： Baddukonda, Tarun Aditya Kanakam, Ronit Kumar Yasotha, B. Srm Institute of Science and Technology Department of Computer Science and Engineering Kaatankulathur Tamilnadu Chennai India School of Computing Srm Institute of Science and Technology Department of Computing Technologies Kaatankulathur Tamilnadu Chennai India

ISBN: (纸本)9798350382693

Adversarial attacks in the Artificial Intelligence have received extensive coverage in the latest years. Such attacks have affected deep learning models and neural networks due to the weaknesses that are part of the AI systems. This exposure has seen classification errors compromises and the undermining of the model performance. Understand the critical mechanisms of adversarial attacks is critical to developing strong defense mechanisms that would protect AI systems from the malevolent interventions. This abstract provides a summary of adversarial attacks as a prevalent threat to the dependability and resilience of AI technologies thus the need for continuous research and development. © 2024 IEEE.

关键词： Adversarial machine learning

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：