检索结果-内蒙古大学图书馆

2024 International Conference on Futuristic Technologies in Control Systems and Renewable Energy, ICFCR 2024

作者： Sundar, Lingala Syam Reddy, N. Abhignya Kumar, N. T. Ravi Prince Mohammad Bin Fahd University Dept. of Mechanical Engineering Al-Khobar31952 Saudi Arabia The University of Texas Department of Computer Science and Engineering Arlington United States Greenko Group Hyderabad India

ISBN: (纸本)9798331504601

The research describes experimentally determine the heat transfer coefficient, Nusselt number, pressure drop, and friction factor of CoFe2O4 /water nanofluids flow in a tube during turbulent flow and then predict them using an artificial neural networkscaled conjugate gradient technique. The nanoparticles are synthesized using chemical reaction method. The water mixed CoFe2O4 hybrid nanofluids were prepared under 0.25% to 1.25% volume loadings. The flow properties such as thermal conductivity, viscosity, density and specific heat were measured in the temperatures from 20 to 60 °C, and volume loadings from 0.25% to 1.25%, followed by heat transfer and friction factor studies. The obtained data is used for ANN-Scaled Conjugate Gradient Predictions. The experimental results are shown that, the Nusselt number and heat transfer coefficient are boosted by 28.26% and 46.79% at φ=1.25% at a Reynolds number of 19301, with a friction factor penalty of 19.04% compared to water data. The results also showed that the scaled conjugate gradient algorithm predicted remarkably accurate values, with correlation coefficients of 0.99383, 0.99528,0.98653, and 0.9826 for heat transfer, Nusselt number, pressure drop, and friction factor, respectively. New Nusselt number, and friction factor correlations are reported based on experimental data. © 2024 IEEE.

关键词： Nusselt number

来源：评论

学校读者我要写书评

暂无评论

A Hybrid Model for Deciphering Doctors' Handwriting Notes Recognition

A Hybrid Model for Deciphering Doctors' Handwriting Notes Re...

引用

IEEE International Conference on Artificial Intelligence in engineering and Technology (IICAIET)

作者： Santosh Khanal Rabindra Bista Dept. of Computer Science and Engineering Kathmandu University Kavre Nepal

ISBN: (数字)9798350389692

ISBN: (纸本)9798350389708

Handwritten Text image Recognition (HTR) in the medical profession has become the major challenge due to the complexity of doctors' handwriting styles and the critical need for precise and efficient text recognition. This study proposes an innovative approach that combines the power of Bidirectional Long Short-Term Memory networks (BLSTM) and Convolutional Neural Networks (CNN) to address these challenges effectively. The proposed model harnesses the bidirectional capabilities of BLSTMs to capture contextual dependencies within doctors' handwritten notes, enabling it to understand and interpret handwriting more accurately. Additionally, CNNs are employed for feature extraction, enabling the model to recognize salient patterns and representations within handwritten text images. This paper presents comprehensive experiments conducted on a diverse dataset of doctors' handwritten notes, demonstrating the model's superior performance compared to conventional approaches and outperforms the existing work in terms of accuracy. Two different crucial metric word error rate and character error rate are accessed through the combination of BLSTM and CNN that yields state-of-the-art accuracy, robustness to variations in handwriting styles, and remarkable adaptability across various medical documents in the context of Nepal.

关键词： Handwriting recognition Adaptation models Accuracy Image recognition Text recognition Error analysis Medical services Feature extraction Convolutional neural networks Biomedical imaging

来源：评论

学校读者我要写书评

暂无评论

Call Translator with Voice Cloning Using Transformers

Call Translator with Voice Cloning Using Transformers

引用

International Conference for Convergence of Technology (I2CT)

作者： Nawaz Fateen Khan Nandan Hemanth Navyae Goyal Pranav KR Pooja Agarwal Computer Science Engineering Dept. PES University Bangalore India

ISBN: (数字)9798350394474

ISBN: (纸本)9798350394481

Voice Cloning employs technology and algorithms to create an artificial or synthetic reproduction of a person’s voice. To understand and mimic the distinct vocal qualities of the target speaker, including tone, pitch, cadence, and pronunciation, a machine learning model must be trained on audio samples of the speaker. This project presents itself as a revolutionary approach to the enduring problem of successful cross-lingual interactions, the paper’s novel approach to speech-to-speech machine translation that incorporates voice cloning. This method allows us to synthesize speech in the target language while maintaining the original speaker’s vocal characteristics. The model algorithms such as Whisper AI(WER 5% for transcription, No-Language-Left-Behind-200(44% more accurate than other models) for translation, Tacotron for speech synthesis and multiple transformers for voice cloning will be incorporated. The Voice Cloning model operates effectively on unseen voices, transcending the limitations of voice cloning will be incorporated. The voice cloning model operates effectively on unseen voices, transcending the limitations of relying solely on pre-trained models. The model successfully clones users’ voices in a different language with an approximate latency of 10 seconds.

关键词： Machine learning algorithms Navigation Cloning Speech recognition Machine learning Linguistics Transformers

来源：评论

学校读者我要写书评

暂无评论

Optimizing CNN Using HPC Tools

arXiv

引用

arXiv 2024年

作者： Rahman, Shahrin Dept. of Computer Science and Engineering Brac University Dhaka Bangladesh

With its vast range of applications, including object recognition, image classification, and pattern recognition, the Convolutional Neural Network (CNN) algorithm has brought about a paradigm shift in the field of computer vision. How- ever, training massive CNN models requires a lot of computational power, which can be efficiently handled by using high- performance computing (HPC) techniques. The CNN method is optimized in this paper using HPC technologies to increase its effectiveness. By utilizing multi-core processors, graphics processing units (GPUs), and parallel computing frameworks like OpenMPI and CUDA, the proposed approach makes use of distributed computing and parallel processing techniques to speed up the training of the CNN model. Using benchmark datasets, a thorough assessment of the optimization strategy is carried out, demonstrating considerable improvements in the CNN algorithm’s performance and training time. Additionally, a comparison of the suggested approach with alternative optimization strategies is conducted, demonstrating its superiority in terms of training time and performance. Overall, this study persuasively demonstrates how HPC technologies may be used to refine the CNN method, resulting in faster and more accurate training of large-scale CNN models. The proposed method has the potential to be more broadly applicable to various deep learning algorithms, which is significant since it will develop effective and efficient machine learning *** Codes N/A © 2024, CC BY-NC-SA.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

Handcrafted Feature Assisted Light-Weight Encoder Decoder Based Classifier for Yoga Posture Recognition

Handcrafted Feature Assisted Light-Weight Encoder Decoder Ba...

引用

Device Intelligence, Computing and Communication Technologies, (DICCT), International Conference on

作者： Prasiddha Sarma S. Ibotombi Singh Dept. of Computer Science and Engineering Tezpur University Tezpur India

ISBN: (数字)9798350372847

ISBN: (纸本)9798350372854

The trend for the need for at-home workout regimens and the potential for technology to support the trend is rising as remote employment becomes more and more common. One such activity that has benefits like increased flexibility and stress alleviation is yoga. Presently, numerous self-guided yoga courses in the form of pictures and videos are available online. Consequently, there is an urgent requirement for a system that can recognize and evaluate the precision of yoga posture execution in these instructional materials. In the current scenario lightweight prediction models are very much essential because devices like smartphones, edge devices, and embedded systems that have limited processing power or battery life, lightweight models are the best option as they require less resources to train and operate. One particular approach of making the recognition system lightweight is the use of keypoint-based pose/action classifiers. In this paper, a lightweight classifier for yoga postures based on neural network architecture using MoveNet pose estimation technique is proposed. Our model is able to classify the yoga postures by learning the human joints information in a efficient manner.

关键词： Pose estimation Neural networks Predictive models Market research Sensors Artificial intelligence Task analysis

来源：评论

学校读者我要写书评

暂无评论

PTdetector: An Automated JavaScript Front-End Library Detector 23

PTdetector: An Automated JavaScript Front-End Library Detect...

引用

Proceedings of the 38th IEEE/ACM International Conference on Automated Software engineering

作者： Xinyue Liu Lukasz Ziarek dept. Computer Science and Engineering University at Buffalo Buffalo USA

ISBN: (纸本)9798350329964

Identifying what front-end library runs on a web page is challenging. Although many mature detectors exist on the market, they suffer from false positives and the inability to detect libraries bundled by packers such as Webpack. Most importantly, the detection features they use are collected from developers' knowledge leading to an inefficient manual workflow and a large number of libraries that the existing detectors cannot detect. This paper introduces PTdetector, which provides the first automated method for generating features and detecting libraries on web pages. We propose a novel data structure, the pTree, which we use as a detection feature. The pTree is well-suited for automation and addresses the limitations of existing detectors. We implement PTdetector as a browser extension and test it on 200 top-traffic websites. Our experiments show that PTdetector can identify packer-bundled libraries, and its detection results outperform existing tools.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A New Approach Towards Deployment and Management of Machine Learning Models Using K-serve Platform for Building Robust Machine Learning Systems 4

A New Approach Towards Deployment and Management of Machine ...

引用

4th IEEE Global Conference for Advancement in Technology, GCAT 2023

作者： Gill, Kanwarpartap Singh Anand, Vatsala Gupta, Rupesh Khosla, Charu Hsiung, Pao-Ann Chitkara University Institute of Engineering and Technology Chitkara University Punjab India Dept. of Computer Science and Information Engineering National Chung Cheng university Chiayi County Taiwan

ISBN: (纸本)9798350305258

Kserve, commonly referred to as KFServing, is an open-source serving framework for machine learning (ML) models that runs on Kubernetes. enables the deployment, servicing, and management of ML models in a production environment in a scalable and effective manner. Kserve resources make it easier to develop scalable and reliable machine learning serving systems by streamlining the deployment and management of ML models in production. With contributions from several organisations and developers in the ML community, it is a dynamic open-source project. The influence that Kserve's work has had on the Data science community is extensively covered in this research. This research also discusses the Kserve design, elements, and features. The additional Kserve platforms that aid in the deployment and administration of machine learning models for creating reliable machine learning systems have also been brought to light. © 2023 IEEE.

关键词： Artificial Intelligence Data science Deep Learning Docker Kserve Kubernetes Machine Learning

来源：评论

学校读者我要写书评

暂无评论

Analytical Survey of Domain Effective Sentiment Lexicon for Aspect-Centric Opinion Analysis

Analytical Survey of Domain Effective Sentiment Lexicon for ...

引用

International Conference on Computing Communication Control and Automation (ICCUBEA)

作者： Lal Babu Purbey Kamlesh Lakhwani Dept. Computer Science of Engineering JECRC University Jaipur India

ISBN: (数字)9798350391770

ISBN: (纸本)9798350391787

Aspect-level sentiment analysis is a modern technology that enables people to share their views or emotions with products, services, or topics based on their aspects. Aspect-level sentiment analysis focuses on aspect extraction, which considers aspects from user reviews. Various methods and approaches have been employed for aspect extraction. Although these methods show better performance and attempt to overcome various challenges, many unsolved problems still need to be solved, such as sentimental ambiguity, opinion misprediction, a weak sentiment lexicon, and implicit sentiment detection. Numerous sentiment lexicons have been employed for word polarity extraction. Dictionaries also play an important role in sentiment classification. There are various dictionaries, but it has been found that issues such as one dictionary might not work properly for all domains. This study analyzes various existing sentiment lexicons employed for domain-specific sentiment analysis. Various approaches have been employed to create a domain-dependent sentiment lexicon and its effect on aspect-level sentiment analyses.

关键词： Surveys Deep learning Sentiment analysis Analytical models Dictionaries Automation Reviews Semantics Context modeling

来源：评论

学校读者我要写书评

暂无评论

Continual Learning by Three-Phase Consolidation

arXiv

引用

arXiv 2024年

作者： Maltoni, Davide Pellegrini, Lorenzo University of Bologna Dept. of Computer Science and Engineering - DISI Italy

TPC (Three-Phase Consolidation) is here introduced as a simple but effective approach to continually learn new classes (and/or instances of known classes) while controlling forgetting of previous knowledge. Each experience (a.k.a. task) is learned in three phases characterized by different rules and learning dynamics, aimed at removing the class-bias problem (due to class unbalancing) and limiting gradient-based corrections to prevent forgetting of underrepresented classes. Several experiments on complex datasets demonstrate its accuracy and efficiency advantages over competitive existing approaches. The algorithm and all the results presented in this paper are fully reproducible thanks to its publication on the Avalanche open framework for continual learning. Copyright © 2024, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Enhancing Vision Models for Text-Heavy Content Understanding and Interaction

arXiv

引用

arXiv 2024年

作者： Adithya, T.G. Adithya, S.K. Bharadwaj, Abhinav R. Abhiram, H.A. Narayan, Surabhi Dept. Computer Science & Engineering PES University Bangalore India

Interacting and understanding with text heavy visual content with multiple images is a major challenge for traditional vision models. This paper is on enhancing vision models' capability to comprehend or understand and learn from images containing a huge amount of textual information from the likes of textbooks and research papers which contain multiple images like graphs, etc and tables in them with different types of axes and scales. The approach involves dataset preprocessing, fine tuning which is by using instructional oriented data and evaluation. We also built a visual chat application integrating CLIP for image encoding and a model from the Massive Text Embedding Benchmark which is developed to consider both textual and visual inputs. An accuracy of 96.71% was obtained. The aim of the project is to increase and also enhance the advance vision models' capabilities in understanding complex visual textual data interconnected data, contributing to multimodal AI. Copyright © 2024, The Authors. All rights reserved.

关键词： Benchmarking

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：