检索结果-内蒙古大学图书馆

Advancements in Speech Recognition: Exploring Feature Extraction Techniques for Enhanced Accuracy and Human-Machine Communication 4

Advancements in Speech Recognition: Exploring Feature Extrac...

引用

4th International Conference on Advances in Electrical, Computing, Communication and Sustainable Technologies, ICAECT 2024

作者： Bhavani, R. Kathirvelu, Kalaivani Vels Institute of Science Technology and Advanced Studies Department of Computer Science and Engineering Chennai India

ISBN: (纸本)9798350343670

Speech Recognition (SR), for all diverse languages, is still a big challenge. SR is predominantly used in telecommunication, medical documentation, car control, aviation, etc. SR involves breaking down an audio signal of spoken words into discrete units and then matching those units to words in a language model. The main target of SR is to attain healthier verbal communication amid humans and machines. Therefore, it is a good technology of the human-machine crossing point. The SR uses the Feature Extraction (FE) techniques for better accuracy. In this paper, various techniques for FE are discussed, and an overview is provided to assist in selecting the most suitable technique for achieving higher accuracy. The review highlights the importance of developing more accurate and robust SR systems to improve the user experience in various domains. Overall, it provides the manuscript deals with five different techniques, and among them, the Mel-Frequency Cepstral Coefficients (MFCC) technique was found to exhibit a high level of accuracy. © 2024 IEEE.

关键词： feature extraction Signal processing Signals Sound Speech recognition

来源：评论

学校读者我要写书评

暂无评论

Traffic Density Prediction System using Deep Learning 2

Traffic Density Prediction System using Deep Learning

引用

2nd International Conference on data science and Information System, ICDSIS 2024

作者： Verma, Hritik Sarkar, Sayan Sadhukhan, Souranil Mishra, Aniket Mondal, Md Ashifuddin Narula Institute Of Technology Department Of Computer Science And Engineering India

ISBN: (纸本)9798350381665

In major cities throughout the world, traffic congestion is still a major problem that negatively affects the efficiency of transportation system. Traffic congestion creates several issues that require a creative and adaptable strategy to solve. This work tackles the problem by accurately forecasting vehicle density by merging state-of-the-art deep learning models, specifically YOLO - v8 for real-time vehicles identification and Stacked Long Short-Term Memory (LSTM) models for forecasting traffic density. These methods are combined to create a powerful vehicle density prediction system. The integration of YOLOv8 with LSTM provides a practical response to the growing issues brought on by traffic congestion in cities. The simulation results shows that accuracy of YOLOv8 is 95.36, while average RMSE value of Stacked LSTM models is 0.591. The proposed model is compared with RNN model for vehicle density forecasting and the proposed model performs better than RNN in terms of accuracy. © 2024 IEEE.

关键词： Traffic congestion

来源：评论

学校读者我要写书评

暂无评论

Early-Stage Diabetes Risk Prediction utilizing Machine Learning with Explainable AI from Polynomial and Binning Feature Generation 2

Early-Stage Diabetes Risk Prediction utilizing Machine Learn...

引用

2nd International Conference on Information and Communication Technology, ICICT 2024

作者： Mamun, Mohammad Chowdhury, Safiul Haque Hussain, Mohammed Ibrahim Iqbal, Md. Sadiq Department of Computer Science and Engineering Bangladesh University Dhaka Bangladesh

ISBN: (纸本)9798331508227

Diabetes is a chronic disease that affects a significant portion of the global population. It occurs when the body cannot produce enough insulin or effectively use the insulin it produces, leading to elevated blood glucose levels. Diabetes is a major contributor to various severe health conditions, including heart disease, stroke, kidney failure, and nerve damage. Early detection of diabetes is crucial in mitigating these associated health risks and improving patient outcomes. In response to the increasing prevalence of diabetes, we have developed an automated system for Early-Stage Diabetes Risk Prediction (ESDRP). This study utilizes a dataset consisting of 16 features from 520 instances. We applied multiple Machine Learning (ML) models, including XGBoost (XGB), Bootstrap Aggregating (BAG), Adaptive Boosting (AdaBoost), Light Gradient Boosted Machine (LGBM), and Gradient Boosting Decision Trees (GBDT), both with and without feature generation techniques. Specifically, we explored polynomial and binning feature generation methods. Our findings indicate that the polynomial feature generation technique combined with XGB yielded the highest performance, achieving an accuracy of 99.22%, precision of 100%, recall of 98.15%, specificity of 99.06%, and F1-score of 100%. Additionally, all the ML models were evaluated using confusion matrices (CM) and ROC curves, with the average performance across 10-fold cross-validation demonstrating robust predictive capabilities. Furthermore, to establish trust in our model's predictions, we incorporated two explainable AI (XAI) methods: LIME and SHAP. These techniques helped us understand feature importance, the decision-making process of the models, and enhanced the reliability of our results. Our automated system aims to assist individuals of all ages and healthcare systems in identifying ESDRP, thereby supporting informed decision-making and improving global health outcomes. ©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

A Distributed Machine Learning Approach to Tertiary Structure Prediction of Protein 15

A Distributed Machine Learning Approach to Tertiary Structur...

引用

15th International Conference on Advances in Computing, Control, and Telecommunication Technologies, ACT 2024

作者： Mahoorkar, Ravi Kiran Niranjan, T.S. Dharshan, H.B. Sivagamasundari, G. Department of Computer Science and Engineering PES University Bengaluru India

ISBN: (纸本)9798331300579

This research investigates the utility of the AlphaFold technique, a deep gaining knowledge of-primarily based method, for the perfect prediction of protein tertiary structures from their corresponding amino acid sequences. The alpha-fold method is a deep learning-based approach that has been shown to accurately predict the 3D structure of proteins with a high degree of accuracy. This involve training the AlphaFold model on a large dataset of known protein structures and their corresponding amino acid sequences, and then using the trained model to predict the structures of novel proteins. The accuracy of the predictions will be evaluated using a variety of metrics, by comparing the reference and predicted structures and Global Distance Test (GDT). The advantages of using the alpha-Fold method, including its high accuracy and wide range of protein targets, as well as its limitations and challenges, such as the need for high computational resources and the limited availability of experimental validation data. The potential applications of accurate protein structure prediction are vast, including aiding in drug discovery, understanding protein function, and guiding experimental studies. The use of the alpha-fold method has the potential to significantly improve the accuracy of protein structure predictions and advance our understanding of the complex world of proteins. © Grenze Scientific Society, 2024.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

A novel relative entropy based FCM algorithm using interval type-2 fuzzy set at local membership functions for brain MR image segmentation 3

A novel relative entropy based FCM algorithm using interval ...

引用

3rd IEEE International Conference on Control, Instrumentation, Energy and Communication, CIEC 2024

作者： Ray, Madhumita Sing, Jamuna Kanta Jadavpur University Computer Science and Engineering Department Kolkata India

ISBN: (纸本)9798350313703

In this proposed article, we develop a fuzzy c-means algorithm based on relative entropy for segmentation of noisy volumetric brain MR images and introduce interval type-2 fuzzy membership functions by using local membership functions as a secondary membership function. The brain magnetic resonance images are influenced by high level of intensity inhomogeneity (IIH) and noise, because of the limitations of machinery devices and cognitive limitations of the operators. As a result it produces blurry images, mostly at the soft tissue boundary regions. This phenomenon increases uncertainty for the image soft tissue classification. The proposed algorithm handle this issue by incorporating interval type-2 fuzzy (IT2F) set at the local membership functions to reduce the underlying uncertainty factors of each voxels more efficiently. It also integrates the total uncertainties for volumetric images by utilizing relative entropy. At first, the proposed algorithm generates the global and local membership functions for each voxel, next, it generates type-2 fuzzy (T2F) set by considering local membership functions as a secondary membership functions and finally, the algorithm generates IT2F set by doing direct defuzzificaton of the T2F set. We conduct qualitative and quantitative experiments on ten different simulated volumes of brain MR images. The results revel the outcome of the experiments and it indicates the supremacy of the proposed algorithm over the state-of-the-art algorithms established in near past. © 2024 IEEE.

关键词： Fuzzy clustering

来源：评论

学校读者我要写书评

暂无评论

LLaMo: Large Language Model-based Molecular Graph Assistant 38

LLaMo: Large Language Model-based Molecular Graph Assistant

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Park, Jinyoung Bae, Minseong Ko, Dohwan Kim, Hyunwoo J. Department of Computer Science and Engineering Korea University Korea Republic of

Large Language Models (LLMs) have demonstrated remarkable generalization and instruction-following capabilities with instruction tuning. The advancements in LLMs and instruction tuning have led to the development of Large Vision-Language Models (LVLMs). However, the competency of the LLMs and instruction tuning have been less explored in the molecular domain. Thus, we propose LLaMo: Large Language Model-based Molecular graph assistant, which is an end-to-end trained large molecular graph language model. To bridge the discrepancy between the language and graph modalities, we present the multi-level graph projector that transforms graph representations into graph tokens by abstracting the output representations of each GNN layer and motif representations with the cross-attention mechanism. We also introduce machine-generated molecular graph instruction data to instruction-tune the large molecular graph-language model for general-purpose molecule and language understanding. Our extensive experiments demonstrate that LLaMo shows the best performance on diverse tasks, such as molecular description generation, property prediction, and IUPAC name prediction. The code of LLaMo is available at https://***/mlvlab/LLaMo. © 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

A Cloud-Based, IoT-Enabled Health Monitoring System Based on Machine Learning Techniques

A Cloud-Based, IoT-Enabled Health Monitoring System Based on...

引用

Congress on Smart Computing Technologies, CSCT 2023

作者： Walia, Mrigana Bathla, Gaurav Department of Computer Science and Engineering Chandigarh University Chandigarh India

ISBN: (纸本)9789819750801

In health care, the use of Cloud-Based and Internet of Things (IoT) technologies has resulted in more efficient and reliable health monitoring systems. The study presents research on a Cloud-Based, IoT-Enabled Health Monitoring System that uses Machine Learning (ML) methods to evaluate and predict health-related data. Several ML methods, such as K-Nearest Neighbors (KNNs), Support Vector Machine (SVM), Random Forest (RF), Logistic Regression (LR), and Naive Bayes (NB), are used to classify and interpret health data. A comparative analysis is done to establish their accuracy and precision rate performance. SVM and RF have the highest accuracy, which is 97.53% and 95.76%, while the precision rates are 95.95% and 95.90% in that order, making them the most effective algorithms for the health monitoring system, respectively. This study contributes to Cloud-Based, IoT-enabled healthcare systems by suggesting how ML can be used to create accurate health monitoring and prediction. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Electronic health record

来源：评论

学校读者我要写书评

暂无评论

Optimized Ensemble Learning based Approach for Heart Disease Prediction 3

Optimized Ensemble Learning based Approach for Heart Disease...

引用

3rd International Conference on Electrical, Electronics, Information and Communication Technologies, ICEEICT 2024

作者： Gera, Sujini Melingi, Sunil Babu Vignan's Foundation for Science Technology and Research Department of Computer Science and Engineering Guntur India

ISBN: (纸本)9798350369083

By using an improved vote classifier model, this study seeks to improve coronary heart disease (CHD) early detection. The study aims to improve the model's accuracy and efficiency by adjusting model parameters and using an enhanced loss function called Focal Loss (FL). The model's performance was assessed using measures including accuracy, precision, recall, and F1-score, utilizing data from the Framingham Heart Study. The voting classifier that was optimized showed remarkable predictive ability, with an accuracy rate of around 98% and almost 100%. These findings point to the model's potential for early cardiac disease identification, which has important ramifications for clinical use. The research highlights how this sophisticated model might lead to better patient outcomes, earlier diagnosis, and lower treatment costs. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Text Augmentation-Based Model for Emotion Recognition Using Transformers

引用

computers, Materials & Continua 2023年第9期76卷 3523-3547页

作者： Fida Mohammad Mukhtaj Khan Safdar Nawaz Khan Marwat Naveed Jan Neelam Gohar Muhammad Bilal Amal Al-Rasheed Department of Computer Science The University of HaripurHaripur22620Pakistan Department of Computer Systems Engineering Faculty of Electrical and Computer EngineeringUniversity of Engineering and Technology PeshawarPeshawar25120Pakistan Department of Electronics Engineering Technology University of TechnologyNowshera24100Pakistan Department of Computer Science Shaheed Benazir BhuttoWomen UniversityPeshawar25000Pakistan Department of Information Systems College of Computer and Information SciencesPrincess Nourah bint Abdulrahman UniversityRiyadh11671Saudi Arabia

Emotion Recognition in Conversations(ERC)is fundamental in creating emotionally ***-BasedNetwork(GBN)models have gained popularity in detecting conversational contexts for ERC ***,their limited ability to collect and acquire contextual information hinders their *** propose a Text Augmentation-based computational model for recognizing emotions using transformers(TA-MERT)to address *** proposed model uses the Multimodal Emotion Lines dataset(MELD),which ensures a balanced representation for recognizing human *** used text augmentation techniques to producemore training data,improving the proposed model’s *** encoders train the deep neural network(DNN)model,especially Bidirectional Encoder(BE)representations that capture both forward and backward contextual *** integration improves the accuracy and robustness of the proposed ***,we present a method for balancing the training dataset by creating enhanced samples from the original *** balancing the dataset across all emotion categories,we can lessen the adverse effects of data imbalance on the accuracy of the proposed *** results on the MELD dataset show that TA-MERT outperforms earlier methods,achieving a weighted F1 score of 62.60%and an accuracy of 64.36%.Overall,the proposed TA-MERT model solves the GBN models’weaknesses in obtaining contextual data for ***-MERT model recognizes human emotions more accurately by employing text augmentation and transformer-based *** balanced dataset and the additional training samples also enhance its *** findings highlight the significance of transformer-based approaches for special emotion recognition in conversations.

关键词： Emotion recognition in conversation graph-based network text augmentation-basedmodel multimodal emotion lines dataset bidirectional encoder representation for transformer

来源：评论

学校读者我要写书评

暂无评论

The Integration of Explainable AI Methods for the Classification of Medical Image data 11

The Integration of Explainable AI Methods for the Classifica...

引用

11th International Conference on Electrical, Electronic and Computing engineering, IcETRAN 2024

作者： Nikolić, Mina Janković, Dragan Stanimirović, Aleksandar Stoimenov, Leonid Faculty of Electronic Engineering Computer Science Department Niš Serbia

ISBN: (纸本)9798350386998

Deep convolutional neural network architectures have in recent years been widely used for enhancing various computer vision tasks, such as Image classification, Semantic Segmentation and Object detection. With great advancements in terms of quality of the obtained results, the path was paved for using these kinds of neural networks in the medical domain. But, when working with sensitive matters involving human lives, there is a need to consider the interpretability and explainability of these models and not just the typical evaluation metrics for the given task. To do such a thing, tools such as LIME and PyTorch Grad-CAM can be used, among many others. The integration of Explainable AI (XAI) methods proposed in this paper aims to enable the paradigm of XAI to be used in medical image classification tasks with the standardized MedMNIST dataset. By doing such an integration, a deeper analysis regarding the quality of the model can be enabled. In that way, instances that were misclassified can be visually examined and used to paint a clearer picture of the complete model's decision-making process. © 2024 IEEE.

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：