检索结果-内蒙古大学图书馆

IEEE Conference on Wireless Communications and Networking

作者： Nargis Fayaz Aman Shreshtha Smruti Sarangi Ranjan K. Mallik Brejesh Lall Dept. of Electrical Engineering Indian Institute of Technology Delhi New Delhi India Dept. of Computer Science and Engg Indian Institute of Technology Delhi New Delhi India

ISBN: (数字)9798350303582

ISBN: (纸本)9798350303599

Semantic communication (SC) aims to convey the meaning of data instead of focusing on its bit-by-bit reconstruction. SC finds applications in beyond 5G and 6G networks for artificial intelligence-empowered multimedia content delivery. In this paper, we propose a novel semantic-aided autoencoder-based image transmission system that leverages semantic information in the form of the segmentation map of an image. We demonstrate up to 23% and 18% improvement (in terms of mean square error and peak signal-to-noise ratio, respectively) in the quality of the received image with only 2% extra bandwidth over a traditional autoencoder-based image transmission system. The study also explores channel coding strategies for our proposed system. We focus on the intrinsically robust nature of semantic data, as compared to traditional data, to design low-density parity check code, Hamming code, and polar code-based unequal error protection (UEP) schemes. Comparative evaluations between UEP and equal error protection schemes show that while both approaches yield similar performance, UEP schemes are more efficient.

关键词： Image segmentation PSNR Image communication Semantics Bandwidth Syntactics Parity check codes

来源：评论

学校读者我要写书评

暂无评论

Vision Transformer Hashing for Image Retrieval

Vision Transformer Hashing for Image Retrieval

引用

2022 IEEE International Conference on Multimedia and Expo, ICME 2022

作者： Dubey, Shiv Ram Singh, Satish Kumar Chu, Wei-Ta Indian Institute of Information Technology Computer Vision and Biometrics Lab Allahabad India National Cheng Kung University Dept. of Computer Science and Information Engineering Taiwan

ISBN: (数字)9781665485630

ISBN: (纸本)9781665485630

Recently, Transformer has emerged as a new architecture in deep learning by utilizing self-attention without convolution. Transformer is also extended to Vision Transformer (ViT) for the visual recognition with a promising performance on ImageNet. In this paper, we propose a Vision Transformer Hashing (VTS) for image retrieval. We utilize the pre-trained ViT on ImageNet as the backbone network and add the hashing head. The proposed VTS model is fine tuned for hashing under six different image retrieval frameworks with their objective functions. We perform the extensive experiments on CIFAR10, ImageNet, NUS-Wide, and COCO datasets. The proposed VTS based image retrieval outperforms the recent state-of-the-art hashing techniques with a significant margin. We also find the proposed VTS model as the backbone network is better than the existing networks, such as AlexNet and ResNet. The code is released at https://***/shivram1987/VisionTransformerHashing. © 2022 IEEE.

关键词： Image retrieval

来源：评论

学校读者我要写书评

暂无评论

Photoplethysmography Signal Quality Assessment Using Neighbour Edge Restricted Horizontal Visibility Graph and Machine Learning Classifiers

Photoplethysmography Signal Quality Assessment Using Neighbo...

引用

International Conference on Electronics, computers and Artificial Intelligence (ECAI)

作者： Zahir Khan M. Sabarimalai Manikandan Mufti Mahmud Dept. of Computer Science and Engineering Indian Institute of Technology Palakkad Palakkad Kerala INDIA Dept. of Electrical Engineering Indian Institute of Technology Palakkad Palakkad Kerala INDIA Dept. of Computer Science Nottingham Trent University Clifton Lane Nottingham UK

ISBN: (数字)9798350371154

ISBN: (纸本)9798350371161

Photoplethysmography (PPG) signals are vital for monitoring pulse rate, blood pressure, and more, but they are prone to motion artefacts and noise, leading to unreliable data. Assessing PPG signal quality is crucial for reliable healthcare and accurate medical diagnoses. By transferring the PPG time domain signal to a Horizontal Visibility Graph (HVG) network and combining extracted features from HVG with machine learning algorithms, we can classify the PPG signal into clean (or high quality) and noisy. We have proposed a new version of HVG called Neighbour Edge Restricted Horizontal Visibility Graph (NERHVG) by invoking some extra conditions for joining edges in HVG for PPG signal quality assessment (SQA). We have used the average degree (AD) of graphs extracted from HVG, and NERHVG algorithms as features in 3 different machine learning classifiers such as Random Forest (RF), Gaussian Naive Bayes (GNB), Decision Tree (DT) to classify 4 standard untrained PPG datasets (DS). The classifier models DT, RF, GNB associated with graph feature AD of HVG, NERHVG algorithms are named as: DT-HVG, DT-NERHVG, RF-HVG, RF-HVG, RF-NERHVG, GNB-HVG and GNB-NERHVG. After all the performance of HVG and NERHVG algorithms using AD feature are compared over the mentioned classifier models. It is observed that the NERHVG algorithm outperformed the HVG algorithm with the AD feature in all 4 datasets with a maximum accuracy of: $99.09\%, 95.03\%, 96.56 \%$ and $84.63\%$ using the GNB classifier.

关键词： Machine learning algorithms Accuracy Gaussian processes Feature extraction Photoplethysmography Classification algorithms Quality assessment

来源：评论

学校读者我要写书评

暂无评论

Image based and Point Cloud based Methods for 3D View Reconstruction in Real-time Environment

Image based and Point Cloud based Methods for 3D View Recons...

引用

IEEE Annual Computing and Communication Workshop and Conference (CCWC)

作者： Arya Agrawal Teena Sharma Nishchal K. Verma Shantaram Vasikarla School of Computer Science and Engineering Vellore Institute of Technology Vellore India Mehta Family School of Data Science and Artificial Intelligence Indian Institute of Technology Guwahati India Dept. of Electrical Engineering Indian Institute of Technology Kanpur India Dept. of Computer Science California State University Northridge CA

3D view reconstruction in a real-time unknown environment poses several challenges to vision-based applications. Herein, this paper outlines real-time experimental results and analysis of 3D view reconstruction methods. Since real-time experimentation requires inputs from different sensors, this paper outlines the results of methods from the following categories: image based methods and point cloud based methods. This paper explains the detailed steps involved in 3D view reconstruction followed by an explanation of the methods. It provides detailed discussions, advantages, limitations, and empirical results, along with presenting differences between algorithms. For analysis of experimental results, two benchmarked datasets i.e., KITTI 3D benchmark suite and ShapeNet are considered. Furthermore, the paper explores potential future scopes, including the utilization of generative models, unsupervised learning, advanced sensors, and many more.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Synchrophasor Measurement Prediction Using Modified Time-Series Mixer Model Under Cyber Attacks

Synchrophasor Measurement Prediction Using Modified Time-Ser...

引用

National Power Systems Conference (NPSC)

作者： Purna Kukadiya Avinash Kumar Pandey Trapti Jain Neminath Hubballi Shailja Shukla Dept. of Electrical Engineering Indian Institute of Technology Indore Indore India Dept. of Electrical Engineering Jabalpur Engineering College Jabalpur India Dept. of Computer Science Engineering Indian Institute of Technology Indore Indore India

ISBN: (数字)9798331519568

ISBN: (纸本)9798331519575

Power systems are critical infrastructures that require robust monitoring and control mechanisms to ensure reliability, stability, and efficiency. It utilizes the data from Phasor Measurement Units (PMUs) and other monitoring devices, which are sent over communication links. Hence, they are subjected to cyber attacks leading to missing or corrupted values. This can cause unreliable operation of the power system. To address the issue, this paper proposes a modified Time-Series Mixer model to accurately predict multivariate measurements under continuous unavailability of data. The proposed model utilizes the time-mixing and feature-mixing layers, which help to capture temporal and spatial correlation to predict the measurements. Further, it can be utilized in real-time applications in case of attack on any communication channel/PMU measurements, as the single model predicts all the PMU measurements. The performance of the proposed model is validated using data generated for the IEEE 14 bus system via RTDS. Numerical results validate the effectiveness of the proposed method, as the errors are significantly smaller than those obtained by the LSTM model-based consecutive PMU measurement prediction technique.

关键词： Power measurement Time series analysis Predictive models Power system stability Phasor measurement units Loss measurement Numerical models Cyberattack Monitoring Mixers

来源：评论

学校读者我要写书评

暂无评论

Data-Driven Assessment of Contaminant Vulnerability in River Ganga Coastal Aquifers

Data-Driven Assessment of Contaminant Vulnerability in River...

引用

OCEANS 2024 - Halifax, OCEANS 2024

作者： Fauziya, Farheen Das, Kousik Majumder, Prakrity Sen, Somdipta Banerjee, Champa Gupta, Ananya Sen Mukherjee, Abhijit Electrical and Computer Engineering University of Iowa Iowa CityIA United States Srm University-AP Department of Env. Science & Engineering Andhra Pradesh India Iit Kharagpur Dept. of Geology and Geophysics West Bengal Kharagpur India School of Environmental Science and Engineering Indian Institute of Technology Kharagpur West Bengal Kharagpur India

ISBN: (纸本)9798331540081

Groundwater of coastal aquifers have strong hy-draulic connections with the coastal hydrodynamics in a various spatiotemporal scale. But the along with the sub daily tidal influence the Tropical cyclone dependent acute groundwater level response has rarely been studied. These acute and instantaneous groundwater level (GWL) fluctuations during tropical cyclones (atmospheric depression) have a direct impact on groundwater flow dynamics and have an impact on instantaneous solute and contaminant mobilization. These groundwater dynamics include influx of sweater and submarine groundwater discharge over the period each atmospheric low-pressure events simultaneously. The dynamic state of GWL fluctuations get stabilized along with decay of atmospheric low-pressure events but the change in solute concentration may stay for a month to seasonal recharge of groundwater and is proportional to the level of acute GWL fluctuations. Thus, this monitoring and prediction of atmospheric low- pressure event dependent GWL fluctuations is a possible indicator of groundwater vulnerability, especially in coastal aquifers. This study conducts a comparative analysis of machine learning and deep learning models to predict groundwater level fluctuations. The models include a Fully Connected Neural Network (FCNN), Artificial Neural Network (ANN), MLPRe-gressor, Support Vector Regressor (SVR), and Random Forest Regressor (RFR). Each model is evaluated using metrics such as Mean Squared Error (MSE), R2 score, Normalized Root Mean Squared Error (NRMSE), Root Mean Squared Error (RMSE), and standard deviation. The results underscore the capability of deep learning techniques to capture nonlinear features in hydrological data, thereby enhancing the understanding and prediction of groundwater levels. Such predictions are crucial for assessing seawater infiltration, which is a potential threat to the available drinking water sources and nutrient and contaminant fluxes within coastal aquifers. Submarine

关键词： Groundwater pollution

来源：评论

学校读者我要写书评

暂无评论

Enhancing Siamese Neural Network Performance with Custom Local Binary Patterns for Face Matching

Enhancing Siamese Neural Network Performance with Custom Loc...

引用

Conference on Information and Communication technology (CICT)

作者： Pawan Singh Koranga Dept. of Computer Science and Engineering Indian Institute of Technology Kharagpur Kharagpur India School of Computer Sciences UPES Dehradun India

ISBN: (数字)9798331531539

ISBN: (纸本)9798331531546

In this study, we investigate the efficacy of incorporating custom local binary patterns (LBP) as a preprocessing technique to enhance the performance of Siamese neural networks for face matching tasks. Our research focuses on training Siamese models using the DC Face dataset and comparing their performance when trained with and without LBP integration. The custom LBP function is designed to capture fine-grained texture details from facial images, providing robust feature representations that are invariant to variations in lighting conditions and facial expressions. Through rigorous experimentation and evaluation, we analyze the impact of LBP on the threshold point for image differentiation and compare the matching accuracy of Siamese models trained with and without LBP. Our findings reveal significant improvements in matching accuracy and robustness when utilizing the texture features extracted by the custom LBP function. By encoding local pixel structures and capturing subtle facial characteristics, LBP-enhanced Siamese models demonstrate enhanced performance in accurately matching faces across different images. Further-more, our study highlights the importance of using domain-specific preprocessing techniques, such as LBP, to optimize the capabilities of Siamese neural networks for face matching tasks. The results underscore the effectiveness of customizing feature extraction methods to suit the characteristics of the dataset, leading to improved matching accuracy and robustness. Overall, our research contributes valuable insights into the role of LBP in enhancing the performance of Siamese neural networks for image-based similarity tasks, with implications for various real-world applications in face recognition and biometric authentication.

关键词： Training Accuracy Image coding Face recognition Neural networks Lighting Feature extraction Robustness Information and communication technology Pattern matching

来源：评论

学校读者我要写书评

暂无评论

Explainability to Business: Demystify Transformer Models with Attention-based Explanations

Explainability to Business: Demystify Transformer Models wit...

引用

Applied Artificial Intelligence and Computing (ICAAIC), International Conference on

作者： Rajasekhar Thiruthuvaraj Ashly Ann Jo Ebin Deni Raj Dept. of Computer Science & Engineering Indian Institute of Information Technology Kottayam India

Recently, many companies are relying on Natural Language Processing (NLP) techniques to understand the text data generated daily. It has become very critical to deal with this data because finding the sentiments of text and summarizing them will help the company understand the pain points of the customers posting reviews on social media or understand the experience of the customer. These requirements have increasingly demanded many advanced algorithms to deal the text data. The introduction of Transformers led to businesses adopting NLP methods more and more to keep up with their needs. Models like Bidirectional Encoder Representations from Transformers (BERT) and Generative Pre-trained Transformers (GPT), state-of-the-art results were achieved with billions of parameters learned. Although these advancements improved the accuracy and expanded the use of algorithms to a wide range of NLP tasks like language translation, text summarization, and language modeling. Businesses are more interested in the Explainability of the model compared to its accuracy. Explainable Artificial Intelligence (XAI) plays an important role to comprehend the complexities of the model as well as the influence of weights on predictions. In this paper, the complexities of the transformer model are unraveled by presenting a straightforward method for computing explainable predictions. The DistilBERT model is chosen as an example to implement the explainable system due to its lighter nature. Combining the strengths of a Posthoc expla-nation with those of a self-learning neural network, the method makes it simple to scale it to other algorithms to implement. With technologies like python, PyTorch, and Hugging Face, a detailed step-by-step algorithmic computation is demonstrated to explain the predictions from the attention-based explanations.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Recurrent Neural Network Based RACH Scheme Minimizing Collisions in 5G and Beyond Networks

Recurrent Neural Network Based RACH Scheme Minimizing Collis...

引用

IEEE Conference on computer Communications Workshops, INFOCOM Wksps

作者： Siba Narayan Swain Ashit Subudhi Dept. of Computer Science and Engineering Indian Institute of Technology Dharwad Karnataka India

Limited preambles in 5G New Radio (NR) can be a bottleneck on the performance of network access procedures. Due to the limited number of preambles, there is a non-zero probability that two mobile User Equipments (UEs) selecting same preamble signatures leading to collisions. Consequently, the base stations (gNBs) in 5G Radio Access Network (RAN) are unable to send a response to the UEs. Furthermore, with the increase in the number of cellular UEs and Machine Type Communication (MTC) devices, the probability of such preamble collisions further increases, thereby leading to reattempts by UEs. This in turn, results in increased latency and reduced channel utilization. In order to reduce contention during preamble access, we propose to use deep learning based models to design a Random Access Channel (RACH) procedure that predicts the incoming connection requests in advance and proactively allocates uplink resources to UEs. We have used Recurrent Neural Network (RNN) which is provided with the history of connection requests to predict UEs which are going to participate in contention based RACH procedure. Finally, we propose a RNN based RACH scheme where the gNB uses RNN model along with the standard RACH process to reduce preamble collisions. On doing extensive simulations, it is observed that there is a significant reduction in the number of collisions when the proposed scheme is employed in a dense user scenario thereby proving the efficacy of the proposed scheme in enabling massive access of users to 5G network.

关键词：

来源：评论

学校读者我要写书评

暂无评论

ARIA: Augmented Reality and Artificial Intelligence enabled mobile application for Yield and grade prediction of tomato crops

引用

Procedia computer science 2024年 235卷 2693-2702页

作者： Balaji Prabhu B V Shashank R Shreyas B Omkar Subbaram Jois Narsipura Dept. of Computer Science and Engineering (AI&ML) Malnad College of Engineering Hassan Visvesvaraya Technological University Belagavi India 573202 Dept. of Computer Science Viterbi School of Engineering University of Southern California Los Angeles USA 90007 Dept. of Computer Science and Engineering RNS Institute of Technology Bangalore India 560098 Dept. of Aerospace Engineering Indian Institute of Science Bangalore India 560012

Agriculture is the most crucial sector of the indian economy. Lately, there has been a surge in the usage of technologies like deep learning and computer vision to make the process of agriculture modern and consequently, lessening the mistakes related to conventional processes. This work delivers a user-friendly, accessible, and novel approach for the detection, counting, and grading of tomatoes found on a farm. An Augmented Reality (AR) based mobile application is developed to obtain the images efficiently from a tomato farm in the pre-harvest stage subjected to open situations. The proposed approach uses Faster RCNN, a convolutional neural network model for detection of tomatoes from the input image on a large scale. The proposed model is trained and tested using 2083 images. The results are then analyzed for the overall performance of detection, segmentation, and classification of tomatoes. The results have examined the efficiency of the proposed mobile application and demonstrate the robustness it exhibits for the detection, grading and yield prediction of Tomatoes.

关键词： tomato detection tomato grading tomato yield prediction augmented reality deep 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：