检索结果-内蒙古大学图书馆

Trends in Quantum computing and Emerging Business Technologies (TQCEBT), International Conference on

作者： S. Pushparani K. Sashi Rekha V. M. Sivagami R. Usharani M. Jothi Department of Computer Science and Engineering Meenakshi College of Engineering Chennai Tamil Nadu Department of Computer Science and Engineering Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Chennai Tamil Nadu India Department of Information Technology Sri Venkateswara College of Engineering Sriperumbudur Chennai Tamil Nadu Department of Computational Intelligence School of Computing Faculty of Engineering and Technology SRM Institute of Science and Technology Kattankulathur Tamil Nadu Department of Artificial Intelligence & Data Science CARE College of Engineering Trichy Tamil Nadu

ISBN: (数字)9798350384277

ISBN: (纸本)9798350384284

Modern audio processing requires compression and restoration for streaming services and data storage and transmission. This study examines how deep learning can address these difficulties and includes an architecture flow diagram. The suggested complex design has numerous integral stages optimized for audio compression and restoration. The input layer contains raw audio data. The pre-processing module normalizes and extracts features from raw data, frequently using Mel-frequency cepstral coefficients (MFCC) to capture audio characteristics. This module improves data quality by reducing noise and filtering. The architecture's heart is the compression module, where deep learning is used. This module compresses audio data using neural networks, maybe CNNs or RNNs. This compression is quantified by compression ratio and bitrate, with hypothetical values representing audio quality trade-offs. The compression module compacts audio data for storage or transmission. To minimize data size, quantization and encoding are common here. To accurately decompress audio, the decompression module mimics the compression module's architecture. Depending on needs, error correction or quality enhancement layers may be added. Filtering, equalization, and dynamic range compression may be used in the post-processing module to improve audio quality. If needed, a restoration module uses deep learning models to minimize noise and artifacts. This study presents four hypothetical scenarios to demonstrate different results. The scenarios emphasize high compression, balanced approaches, quality prioritization, and low resource utilization. Deep learning approaches can adapt to varied audio processing needs by balancing compression ratio, bitrate, audio quality (measured by PSNR), and restoration accuracy in each case.

关键词： Deep learning Training Filtering Bit rate Noise Computer architecture Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Methodical Tamil Character Recognition Using Fabricated CNN Model

Methodical Tamil Character Recognition Using Fabricated CNN ...

引用

International Conference on Computer Communication and Informatics (ICCCI)

作者： M Bhavani S Ravikumar S. Prithi Y Arockia Raj Babu Rajendiran Department of Computer Science and Engineering Rajalakshmi Engineering College Chennai Tamilnadu India Department of Computer Science and Engineering Vel Tech Rangarajan Dr.Sagunthala R&D Institute of Science and Technology Chennai Department of Computational Intelligence School of Computing SRM Institute of Science and Technology Chennai Department of Computer Science and Engineering PSNA College of Engineering and Technology Dindigul

Recognizing text is one of the most challenging problems in computer vision. To put it simply, handwritten character recognition is the process of recognising text in photographs, papers, and other media and converting it into a form that a computer can read. You can solve this issue by employing the image classification method. We will be able to recognise the user’s typed Tamil characters in this paper. We designed a brand new model, MVGG16, to handle character recognition and classification. Its basic structure is analogous to that of VGG16, but we added a few more layers to improve accuracy. So far, 156 Tamil characters have been learnt, with 350 images each class. In experimental settings, this model achieves an accuracy of 91.8%. As an input method, this work makes use of a whiteboard where the user can scribble the Tamil character to be recognised. The character will be sent for verification after the user has written it and the image has been exported. Once the model identifies the character, it will show it to the user.

关键词： Training Handwriting recognition Temperature Text recognition Urban areas Predictive models Media

来源：评论

学校读者我要写书评

暂无评论

MLRegTest: a benchmark for the machine learning of regular languages

The Journal of Machine Learning Research

引用

The Journal of Machine Learning Research 2024年第1期25卷 13641-13685页

作者： Sam Van der Poel Dakotah Lambert Kalina Kostyszyn Tiantian Gao Rahul Verma Derek Andersen Joanne Chau Emily Peterson Cody St. Clair Paul Fodor Chihiro Shibata Jeffrey Heinz School of Mathematics Georgia Institute of Technology Department of Computer Science Haverford College Department of Linguistics & Institute of Advanced Computational Science Stony Brook University Department of Computer Science Stony Brook University Department of Computer Science Stony Brook University Department of Linguistics Stony Brook University Department of Advanced Sciences Graduate School of Science and Engineering Hosei University

Synthetic datasets constructed from formal languages allow fine-grained examination of the learning and generalization capabilities of machine learning systems for sequence classification. This article presents a new benchmark for machine learning systems on sequence classification called MLRegTest, which contains training, development, and test sets from 1,800 regular *** kinds of formal languages represent different kinds of long-distance dependencies, and correctly identifying long-distance dependencies in sequences is a known challenge for ML systems to generalize successfully. MLRegTest organizes its languages according to their logical complexity (monadic second order, first order, propositional, or restricted propositional) and the kind of logical literals (string, tier-string, subsequence, or combinations thereof). The logical complexity and choice of literal provides a systematic way to understand different kinds of long-distance dependencies in regular languages, and therefore to understand the capacities of different ML systems to learn such long-distance ***, the performance of different neural networks (simple RNN, LSTM, GRU, transformer) on MLRegTest is examined. The main conclusion is that performance depends significantly on the kind of test set, the class of language, and the neural network architecture.

关键词： formal languages regular languages subregular languages sequence classification neural networks long-distance dependencies

来源：评论

学校读者我要写书评

暂无评论

Early Detection of Alzheimer's Disease using Vision Transformer based Convolutional Neural Network

Early Detection of Alzheimer's Disease using Vision Transfor...

引用

Distributed computing and Electrical Circuits and Electronics (ICDCECE), IEEE International Conference on

作者： S Aruna Priyadarshini R Taki Alrwbaye PR Buvaneswari KG Parthiban Department of Computational Intelligence School of Computing SRM Institute of Science and Technology India Department of Information Science and Engineering Nitte Meenakshi Institue of Technology Bangalore India Department of Computers Techniques Engineering College of Technical Engineering The Islamic University Najaf Iraq School of Electrical and Electronics Engineering VIT Bhopal University Sehore India Department of Biomedical Engineering Dhaanish Ahmed Institute of Technology Coimbatore India

ISBN: (数字)9798331533663

ISBN: (纸本)9798331533670

In recent scenarios, Alzheimer’s Disease (AD) has posed early recognition challenges due to complex and overlapping symptoms by making difficulties in timely care and intervention. The existing models such as Convolutional Neural Networks (CNN) struggled with limited interpretability. In this research, Vision Transformer based CNN (ViT-CNN) is proposed for AD diagnosis. The proposed ViT-CNN utilizes query-attention mechanism which mainly focuses on the specific region of input that improves the interpretability issues. This procedure begins with the collection of input images from AD Neuroimaging Initiative (ADNI) dataset. Then, the pixel intensities are normalized and data augmentations such as flipping is performed to increase diversity of input data. After that, ViT is incorporated to extract features by capturing the spatial as well as temporal relationships within the image. Finally, the extracted features are further processed into the CNN model for early diagnosis of AD detection. From the results, proposed ViT-CNN outperformed the existing CNN in terms of accuracy (0.83), specificity (0.79) sensitivity (0.85), and F1-score (0.85) respectively.

关键词： Neuroimaging Computer vision computational modeling Feature extraction Transformers Data augmentation Data models Convolutional neural networks Alzheimer's disease Integrated circuit modeling

来源：评论

学校读者我要写书评

暂无评论

IoT-Enhanced Path Planning and Obstacle Avoidance for Autonomous Mobile Robots

IoT-Enhanced Path Planning and Obstacle Avoidance for Autono...

引用

Advancement in technology (ICONAT), International Conference for

作者： N Selvanathan Sr Ramya C Sureshkumar R. Kiruthiga J Akshya Mani Deepak Choudhry M. Sundarrajan Department of Information Technology Sona College of Technology Salem Department of Computer Science and Engineering Dr NGP Institute of Technology Coimbatore Department of Information Technology KGiSL Institute of Technology Coimbatore Department of AI&DS Bannari Amman Institute of Technology Erode Department of Computational Intelligence SRM Institute of Science and Technology Chennai Department of Computing Technologies SRM Institute of Science and Technology Chennai Department of Networking and Communications SRM Institute of Science and Technology Chennai

ISBN: (数字)9798350354171

ISBN: (纸本)9798350354188

Path planning and obstacle avoidance play a key role in how well autonomous mobile robots work and stay safe when they move through tricky areas. Traditional path-planning methods often can’t keep up when new obstacles pop up or existing ones change, which makes them less useful in the real world. To fix these problems, the proposed research work brings in a new IoT-powered path planning and obstacle avoidance system that uses the $\mathrm{D}^{*}$ Lite and Deep Reinforcement Learning (DRL) algorithms. Our approach brings together D* Lite’s quick replanning skills with DRL’s ability to learn and adapt made even better by live data from IoT sensors. We’ve put our model through its paces in different simulated environments set up to be as tricky as real-world navigation. Quantitative analysis shows it’s a big step up from older models. Adding IoT data leads to a 30% boost in how paths are planned and cut the time robots need to get through complex routes by 20%. Plus, using DRL helps robots deal with moving obstacles 25% better, which means fewer potential crashes. This is a big step forward in the field of autonomous robotics. It gives a robust answer that would not just be reactive to the given environmental inputs but also, through proactivity, prompt possible obstacles, further contributing to smoother and safer output. As we move along with the development of our methodology, we foresee broad applications in many sectors enhancing the autonomy and efficiency of robotic systems working in dynamic and unpredictable environments.

关键词： Measurement Adaptation models Navigation Statistical analysis Heuristic algorithms Robot sensing systems Path planning Mobile robots Collision avoidance Autonomous robots

来源：评论

学校读者我要写书评

暂无评论

Detection of Military Vehicles based on Edge Devices from Unmanned Aerial Vehicle

Detection of Military Vehicles based on Edge Devices from Un...

引用

Distributed computing and Optimization Techniques (ICDCOT), International Conference on

作者： L Umasankar Anitha D S Trisheela Hassan M. Al-Jawahry Subba Reddy Borra Electrical and Electronics Engineering S.A.Engineering College Poonamallee Tiruvallur India Department of Computational Intelligence School of computing SRM Institute of Science and Technology Kattankulathur Tamilnadu India Dept of CSE Nitte Meenakshi Institute of Technology Bengaluru India The Islamic university Najaf Iraq Computer Science And Engineering Malla Reddy Engineering College for Women (Ugc-Autonomous) Maisammaguda Hyderabad India

The recent technology in surveillance for military vehicle parts forms the basis of intelligence warfare through information realized through monitoring and tracking operations in complex situations. Therefore, the process of identifying and classifying vehicles by an aerial vehicle installed with a resource-limited device and an intelligent object detection algorithm significantly assists security agencies. It is possible to make use of this vehicle either manually or autonomously. To overcome these issues, edge device based military vehicle detection from Unmanned Aerial Vehicle (UAV) is proposed. At present, no publicly available dataset exists that includes different military classes. Therefore, the proposed approach use a dataset consisting of 6772 images labeled as Tanks, Military Trucks, Helicopters, Aircraft, Civilian Cars and Civilian Aircraft. This data set is used to train deep learning models such as Quantized SSD Mobilenetv2 and Tiny Yolo v5 that are then compared with resource limited edge devices. The results reveal that Tiny Yolo v5 outperforms other models, showing high efficiency and suitability for edge-based devices due to its lightweight design.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Tuning 2D magnetism in Fe3+XGeTe2 films by element doping

引用

National science Review 2022年第6期9卷 163-170页

作者： Shanshan Liu Zihan Li Ke Yang Enze Zhang Awadhesh Narayan Xiaoqian Zhang Jiayi Zhu Wenqing Liu Zhiming Liao Masaki Kudo Takaaki Toriyama Yunkun Yang Qiang Li Linfeng Ai Ce Huang Jiabao Sun Xiaojiao Guo Wenzhong Bao Qingsong Deng Yanhui Chen Lifeng Yin Jian Shen Xiaodong Han Syo Matsumura Jin Zou Yongbing Xu Xiaodong Xu Hua Wu Faxian Xiu State Key Laboratory of Surface Physics and Department of Physics Fudan University Institute for Nanoelectronic Devices and Quantum Computing Fudan University College of Science University of Shanghai for Science and Technology Laboratory for Computational Physical Sciences(MOE) Fudan University Solid State and Structural Chemistry Unit Indian Institute of Science School of Electronic Science and Engineering Nanjing University Department of Physics University of Washington Department of Electronic Engineering Royal Holloway University of London Materials Engineering The University of Queensland Beijing Key Laboratory of Microstructure and Property of Advanced Materials Institute of Microstructure and Property of Advanced Materials Beijing University of Technology The Ultramicroscopy Research Center Kyushu University State Key Laboratory of ASIC and System School of Microelectronics Fudan University Collaborative Innovation Center of Advanced Microstructures Department of Applied Quantum Physics and Nuclear Engineering Kyushu University Centre for Microscopy and Microanalysis The University of Queensland Shanghai Research Center for Quantum Sciences

Two-dimensional（2D） ferromagnetic materials have been discovered with tunable magnetism and orbital-driven nodal-line features. Controlling the 2D magnetism in exfoliated nanoflakes via electric/magnetic fields enables a boosted Curie temperature（TC） or phase transitions. One of the challenges, however, is the realization of high TC2D magnets that are tunable, robust and suitable for large scale fabrication. Here, we report molecular-beam epitaxy growth of wafer-scale Fe3+XGeTe2films with TCabove room temperature. By controlling the Fe composition in Fe3+XGeTe2, a continuously modulated TCin a broad range of 185–320 K has been achieved. This widely tunable TCis attributed to the doped interlayer Fe that provides a 40% enhancement around the optimal composition X = 2. We further fabricated magnetic tunneling junction device arrays that exhibit clear tunneling signals. Our results show an effective and reliable approach, i.e. element doping, to producing robust and tunable ferromagnetism beyond room temperature in a large-scale 2D Fe3+XGeTe2fashion.

关键词： 2D ferromagnetic material Fe3+XGeTe2 film element doping above room temperature TC tunability

来源：评论

学校读者我要写书评

暂无评论

Optimizing Routing Paths in Mobile Wireless Sensor Networks: A Sub-Flow Adaptive Multipath Approach for Energy Efficiency and Delay Sensitivity

Optimizing Routing Paths in Mobile Wireless Sensor Networks:...

引用

International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC)

作者： R. Vinston Raja Shavej Ali Siddiqui Avilasha BG L. Mohana Kannan Department of Computational Intelligence Faculty of Engineering and Technology School of Computing SRM Institute of Science and Technology Chennai Tamil Nadu India Deparment of Applied Science and Humanities Faculty of Engineering and Technology Khwaja Moinuddin Chishti Language University Lucknow India Deparment of Mechanical Engineering Dayanandasagar College of Engineering Bangalore India Department of Biomedical Engineering Erode Sengunthar Engineering College Perundurai Tamil Nadu India

ISBN: (数字)9798350376425

ISBN: (纸本)9798350376432

Wireless Sensor Networks (WSNs) have become crucial in various domains such as social applications, healthcare, and the military. However, their diversity makes them susceptible to a wide range of attacks, leading to challenges like poor throughput, high network latency, and excessive energy consumption. This study proposes a novel approach that combines residual Riemannian neural networks with the Sub-Flow Adaptive Multipath Routing algorithm for enhanced routing and attack detection in WSNs. By leveraging node position data, the Sub-Flow Adaptive Multipath Routing algorithm optimizes energy consumption and extends network lifespan. Simultaneously, residual Riemannian neural networks analyze residual patterns in network traffic data to accurately detect rogue nodes, thereby enhancing security. Blockchain technology is then employed to securely store classified data, ensuring data integrity and security. This integrated approach significantly improves network performance, maintains robust security against potential threats, and enhances the reliability of WSNs in demanding environments.

关键词： Wireless sensor networks Energy consumption Adaptive systems Neural networks Routing Throughput Delays Blockchains Security Reliability

来源：评论

学校读者我要写书评

暂无评论

Effective Surrogate Models for Docking Scores Prediction of Candidate Drug Molecules on SARS-CoV-2 Protein Targets

Effective Surrogate Models for Docking Scores Prediction of ...

引用

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Wenqi Shi Mio Murakoso Xiaoyan Guo Linxi Xiong Matthew Chen May D. Wang Department of Electrical and Computer Engineering Georgia Institute of Technology Atlanta USA Department of Applied Life Science Kyoto University Kyoto Japan College of Computing Georgia Institute of Technology Atlanta USA Tanwei College Tsinghua University Beijing China Wallace H. Coulter Department of Biomedical Engineering Georgia Institute of Technology Atlanta USA

Emerging infectious diseases, such as coronavirus disease 2019 (COVID-19), pose a major threat to public health and present a critical challenge for drug discovery. Due to the cost- and time-consuming process of new drug development, virtual pre-screening methods such as protein-ligand docking prediction have become essential tools in enhancing drug refurbishment and repurposing. In this study, we propose a machine learning-based surrogate model for docking score prediction of drug candidates on SARS-CoV-2 protein targets via deep feature concatenation. We investigate 14 different combinations of rule-based and data-driven fingerprinting methods to identify the optimal representation of candidate drug molecules. Extensive experiments on docking scores of 270,000 molecules across 18 different SARS-CoV-2 protein targets demonstrate the effectiveness of the proposed surrogate models. In addition to unseen drugs, we further investigate the generalization of the proposed framework for unseen protein targets. This study may provide an instrumental and generalizable framework for exploring ligand-protein interaction, serving as a useful tool to facilitate rapid drug pre-screening during emerging public health crises.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Enhancing Medical Image Diagnosis with Convolutional Neural Networks

Enhancing Medical Image Diagnosis with Convolutional Neural ...

引用

International Conference on Advances in Information technology (ICAIT)

作者： K. Mahadevan S. Pushparani K. SashiRekha V.M. Sivagami R. Usharani Department of Computer Science and Engineering CARE College of Engineering Trichy Tamil Nadu India Department of Computer Science and Engineering Meenakshi College of Engineering Chennai Tamil Nadu India Department of Computer Science and Engineering Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Chennai Tamil Nadu India Department of Information Technology Sri Venkateswara College of Engineering Chennai Tamil Nadu India Department of Computational Intelligence Faculty of Engineering and Technology School of Computing SRM Institute of Science and Technology Kattankulathur Chennai India

ISBN: (数字)9798350383867

ISBN: (纸本)9798350383874

This paper is a comparative analysis of medical image diagnosis algorithms with Convolutional Neural Networks (CNNs) and other methods; such as Support Vector Machine (SVM), Random Forest, and k-nearest Neighbors (k-NN) The study performs evaluations on datasets with multi-modalities like medical images and diagnostic and various imaging modalities. Results shows that CNNs have the upper hand over traditional, static algorithms across several metrics that assesses the performance of the system such as accuracy, precision, recall, and F1. In particular, CNN precision was 90.5% while the recall was 93.7%, and the accuracy was 92.1% as well as F1-score 92.0%. In addition, the high area under the Receiver Operating Characteristic curve (AUC-ROC) of over 0.95 for the CNN also signed for its excellent discriminatory power. These outcomes only exemplify CNN usefulness in the diagnostic area because it is an exact and accurate method for analyzing medical images. Although the nets CNNs are deployed in clinical cases may help address concerns regarding interpretability, compliance with regulations, and ethical opinions, such issues still exist and require further exploration. Overall, this research takes the first steps in directing algorithmic approach towards medical diagnosis and show the feasibility in deployment of CNNs as inseparable pieces of equipment in modern healthcare.

关键词： Support vector machines Ethics Accuracy Recurrent neural networks Machine learning algorithms Nearest neighbor methods 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：