检索结果-内蒙古大学图书馆

Multi-channel CNN to classify Nepali COVID-19 related tweets using hybrid features

Journal of Ambient Intelligence and Humanized Computing 2024年第3期15卷 2047-2056页

作者： Sitaula, Chiranjibi Shahi, Tej Bahadur Department of Electrical and Computer Systems Engineering Monash University Wellington Rd ClaytonVIC3800 Australia Earth Observation and AI Research Group Department of Infrastructure Engineering The University of Melbourne ParkvilleVIC3010 Australia School of Engineering and Technology Central Queensland University Rockhampton Australia Central Department of Computer Science and Information Technology Tribhuvan University TU Rd Kathmandu Kirtipur44618 Nepal

Because of the current COVID-19 pandemic’s increasing fears among people, it has triggered several health complications such as depression and anxiety. Such complications have not only affected developed countries but also developing countries such as Nepal. These complications can be understood from peoples’ tweets and comments posted online after their felicitous analysis and sentiment classification. Nevertheless, owing to the limited number of tokens and words in each tweet, it is crucial to capture multiple pieces of information associated with them for better understanding. In this study, we, first, represent each tweet by combining both syntactic and semantic information, called hybrid features. The syntactic information is generated from the bag-of-words method, whereas the semantic information is generated from the combination of the fastText-based (ft) and domain-specific (ds) methods. Second, we design a novel multi-channel CNN (MCNN) model, which ensembles the multiple CNNs, to capture multi-scale information for better classification. Last, we evaluate the efficacy of both the proposed feature extraction method and the MCNN model classifying tweets into three sentiment classes (positive, neutral, and negative) on the NepCOV19Tweets dataset, which is the only public COVID-19 tweets dataset in the Nepali language. The evaluation results show that the proposed hybrid features outperform individual feature extraction methods with the highest classification accuracy of 69.7% and the MCNN model outperforms the existing methods with the highest classification accuracy of 71.3% during classification. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2023.

关键词： COVID-19

来源：评论

学校读者我要写书评

暂无评论

Detection of underwater objects using Deep Learning and You Look Only Once Algorithm

Detection of underwater objects using Deep Learning and You ...

引用

Mysore Sub Section International Conference (MysuruCon), IEEE

作者： Korutla Meghana Rudraraju Vandana Padavala Vinod Kumar Polisetty Swetha M. A. Jabbar Computer Science and Engineering (AI&ML) Vardhaman College of Engineering Shamshabad India

Underwater object detection plays a pivotal role in various applications ranging from marine biology research to underwater exploration and surveillance. This paper presents a novel approach to underwater object detection using the YOLOV5 algorithm. The proposed underwater object detection model leverages the capabilities to efficiently and accurately identify and localize objects in underwater imagery. We used the aqua pretrain dataset for training and detecting distinct fish species in different water bodies and achieved highest mean average precision value of 0.861 for jelly fish class. This proposed method is helpful in identifying the debris or structures located in the depth of oceans and dealing with underwater calamities.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Facial Expression Detection Using Recurrent Residual Convolutional Neural Network

Facial Expression Detection Using Recurrent Residual Convolu...

引用

Mysore Sub Section International Conference (MysuruCon), IEEE

作者： Lingaladinne Yoshitha Reddy Prakash Kumar Saranghi Chakka Manaswini M.A. Jabbar Suram Abhiram Reddy Computer Science Engineering(AI&ML) Vardhaman College of Engineering Shamshabad Hyderabad

In several industries, such as healthcare, emotion analysis, and human-computer interface, facial expression detection is crucial. Recurrent residual convolutional neural networks (RRCNNs) are a revolutionary technique used in this to identify facial expressions. Our suggested model uses residual learning and recurrent connections to improve the precision and reliability of facial emotion recognition. This study is based on the incorporation of recurrent connections like ResNet into a sophisticated residual CNN architecture. This combination enables our model to extract from facial sequences both spatial and temporal information, making it an effective tool for tracking changing facial expressions over time. To train and evaluate our model, we used well- known facial expression datasets such as *** face expression identification model we've suggested, which parallelizes residual CNNs and recurrent neural networks, provides a potential method for enhancing the accuracy of 61.86% and completeness of emotion recognition systems. The RRCNN has a significant advantage in facial expression detection, even with a less accuracy. Due to its distinct architecture, the model excels at tasks like video-based detection where expressions change over time. The potential of the model lies in its ability to learn subtle temporal patterns, even though the accuracy may not be as good as it could be with more data, hyperparameter tuning, and model refinement. Because of its innate ability to analyze data sequentially, it can be a useful tool in the quest for context-aware and more precisefacial expression recognition and it is perfect for a range of real-world applications, including virtual reality, human-computer interaction, and mental health monitoring.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Applying Multi-Layer Perceptrons (mlP) for Demand Forecasting in Supply Chain Management 2

Applying Multi-Layer Perceptrons (MLP) for Demand Forecastin...

引用

2nd IEEE International Conference on IoT, Communication and Automation Technology, ICICAT 2024

作者： Ankam, Sreejyothsna Durgadevi, G. Kumar, Rakesh Mounika, Bathula Alsalami, Zaid Niranjanamurthy, M. GRIET Bachupally Department of CSBS Telangana Hyderabad India New Prince Shri Bhavani College of Engineering and Technology Department of ECE Tamil Nadu Chennai600073 India GLA University Department of Computer Engineering & Applications Mathura India CMR Institute of Technology Department of Computer Science and Engineering Telangana Hyderabad India College of Technical Engineering Department of Computers Techniques Engineering The Islamic University Najaf Iraq BMS Institute of Technology and Management Department of AI and ML Karnataka Bangalore560064 India

ISBN: (纸本)9798350368109

Demand forecasting plays a significant role in supply chain management and business operational effectiveness. This paper focuses on the use of Multi-Layer Perceptrons (mlP) in demand forecasting which is one of artificial neural networks due to its proved ability to model non-linear patterns. This typology suggests a detailed approach which embraces data gathering, data preparation, model building and model assessment. The mlP model is compared with other models including ARIMA, Random Forest and LSTM models. Therefore, the use of mlP model indicated improved performance in terms of RMSE and MAPE in comparison to traditional approaches. Although LSTM has slightly higher performance, it is relatively more simpler and faster to implement mlP which results to better accuracy and fewer computations. It is identified that historical sales data, seasonal factors and economic factors are some of the most important features that affect the accuracy of the forecast. The use of mlP during the implementation of the project proves the efficiency of the proposed approach and the obtained results. This paper reveals the possibility of mlPs in improving the part of demand forecasting in the supply chain management process and suggested ways that could increase opportunities for better supply chain management through the management of inventory. Further research can be dedicated to the elaboration of the combined approaches utilizing the advantages of the notable methodologies for sales forecasting.. © 2024 IEEE.

关键词： Demand response

来源：评论

学校读者我要写书评

暂无评论

Harnessing Deep Learning Models in Fundus Imaging for Early Diagnosis in Eye Disorders

Harnessing Deep Learning Models in Fundus Imaging for Early ...

引用

Smart Technologies, Communication and Robotics (STCR)

作者： E.Priya M.Dhinesh G.Rohitkumar M.Dheena Department of Computer Science and Engineering(AI&ML) Sri Sairam Engineering College Chennai India Department of Electronics and Communication Sri Sairam Engineering College Chennai India

ISBN: (数字)9798350357530

ISBN: (纸本)9798350357547

Throughout the globe, eye conditions such as diabetic retinopathy, glaucoma, cataract, and age-related macular degeneration are the major causes of blindness and visual impairment. Early detection is crucial to prevent permanent loss of vision. Fundus imaging is important for early detection of these disorders because it produces detailed images of the retina. The project aims to develop an automated system that can detect various eye disorders from these fundus images using deep learning. To build the model, a diverse collection of labeled fundus images was sourced from openly available platforms like Kaggle, EyePACS, and Messidor. These images were carefully prepared before training through processes like resizing, enhancing contrast, and applying basic transformations. The system employs a locally developed Convolutional Neural Network along with transfer learning-based ResNet50. The CNN model is designed to gradually extract important features from the images using standard building blocks like activation functions and pooling techniques, while also including methods to improve stability and prevent overfitting. The model was developed using the Django framework, along with user access to real-time diagnostic feedback through a web-based interface.

关键词： Training Macular degeneration Glaucoma Diabetic retinopathy Feature extraction Stability analysis Real-time systems Convolutional neural networks Residual neural networks Overfitting

来源：评论

学校读者我要写书评

暂无评论

Intrusion Detection System Using Improved Gloden Jackal optimization for Security and Privacy in Internet of Things 2

Intrusion Detection System Using Improved Gloden Jackal opti...

引用

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

作者： Bhagavatham, Naresh Kumar Al-Farouni, Mohammed Kumar, Komuravelly Sudheer Sundaram, S. Meenakshi Puneeth, R.P. Vignana Bharathi Institute Of Technology Dept Of Computer Science And Engineering Hyderabad India The Islamic University College Of Technical Engineering Department Of Computers Techniques Engineering Najaf Iraq Sr University School Of Computer Science & Artificial Intelligence Telangana Warangal India Nitte Meenakshi Institute Of Technology Department Of Ai&ds Bengaluru India Department Of Computer Science And Engineering Karnataka India

ISBN: (纸本)9798350381665

Technology plays a primary role in rapid growth of service and identifying the quality of life. Recent technology such as Internet of Things (IoT) determines an impressive performance in the development of fast-forward. Intrusion Detection System (IDS) is employed as a lifeline to prevent attacks by categorizing activities as suspicious or normal. However, identifying the most relevant features from large data is challenging to effectively distinguish among suspicious or normal behavior. This research proposes the Improved Gloden Jackal optimization (IGJO) to select the appropriate features in IDS for security and privacy. GJO is improved by using tent mapping reverse learning for initializing populations, sine and cosine techniques are performed for updating prey positions which increases the ability of global exploration, and Cauchy mutation is used to acquire the optimal solutions. Initially, the CIC-IDS2017 dataset is used to evaluate the model performance and min-max normalization is utilized for normalizing the data. Then, IGJO is performed to select the appropriate features and Long Short-Term Memory (LSTM) is employed to classify IDS. The IGJO achieves a high accuracy of 99.95% compared to existing techniques respectively. © 2024 IEEE.

关键词： Internet of things

来源：评论

学校读者我要写书评

暂无评论

Hybrid Deep Learning Model for Spectrum Sensing Classification in Cognitive Radio Networks

Hybrid Deep Learning Model for Spectrum Sensing Classificati...

引用

2024 International Conference on Distributed Systems, computer Networks and Cybersecurity, ICDSCNC 2024

作者： Padmakala, S. Alsalami, Zaid Supriya, R. Sahana, M.P. Deepthi, Palakurthy Saveetha University Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Department of Computer Science and Engineering Chennai India The Islamic University College of Technical Engineering Department of Computers Techniques Engineering Najaf Iraq Nitte Meenakshi Institute of Technology Department of Ise Bengaluru India Dayananadha Sagar College of Engineering Department of Cse Bangalore India Gokaraju Rangaraju Institute of Engineering and Technology Department of Ai&ml Hyderabad India

ISBN: (纸本)9798350375442

Spectrum sensing (SS) uses cognitive radio technology at base transceiver stations in order to discover the licensed spectra used by Primary Users (PUs). This enables the SUs to make wireless network transmissions without causing interference. Cognitive Radio Networks (CRNs) help in the problem on the restricted radio frequency spectrum that wireless devices depend on. For enhanced classification of SS in CRN, the proposed research focused on using hybrid Deep Learning (DL) model by integrating Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM). Initially, the data is generated through eight modulation types considering the varied oversampling rate. The spatial features of signal are extracted through CNN and temporal features are identified by LSTM. the Cooperative Spectrum Sensing (CSS) algorithm is integrated increase the reliability of spectrum detection. The hyper parameters of the trained model are fine tuned for enhanced classification of SS. Experimental results showed that the proposed LSTM-CNN-CSS achieved sensing accuracy at 98.34%, precision at 95.62%, recall at 96.77% when compared to the traditional SS classification models such as Ensemble Convolutional Recurrent Neural Network (ECRNN) and LSTM-Extreme Learning Machine (LSTM-ELM). © 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

Revolutionizing Healthcare with Wearable Devices and IoT Integration for Personalized Health Management in Next-Generation Health Informatics

Revolutionizing Healthcare with Wearable Devices and IoT Int...

引用

2024 IEEE International Conference on Recent Advances in science and engineering Technology, ICRASET 2024

作者： Parthipan, V. Ramesh, P.S. Gururama Senthilvel, P. Dineshkumar, R. Vennila, G. Saveetha Institute of Medical and Technical Sciences Saveetha University Saveetha School of Engineering Department of Computer Science and Engineering Chennai India Vel Tech Rangarajan Dr. Sagunthala R&d Institute of Science and Technology Tamil Nadu Chennai India Saveetha University Saveetha School of Engineering Department of Computer Science and Engineering Chennai India Saveetha Institute of Medical and Technical Sciences Saveetha University Saveetha School of Engineering Department of Electronics and Communication Engineering Chennai India Mohan Babu University School of Computing Department of Ai & Ml Andhra Pradesh Tirupati India

ISBN: (纸本)9798350388602

This work outlines a paradigm shift that is created by the integration of wearable devices with the Internet of Things for health consciousness. With the help of the proposed system, the user is given the constant readings of the key heath indicators including heart rate, blood pressure, blood glucose level, physical activity, and the quality of sleep, all the data are transferred in real-time to an IoT platform, and the data analysis supported by machine learning algorithms provide the user with individually tailored heath reports. Wearable Medical today carried out a 90 persons trial over a six-month long program and noted changes in health parameters, user interactions and results. Several client outcomes highlighted that clients' heart rate, blood pressure and glucose levels, physical activity, and sleep quality improved. The system also provided strong control in data protection and the compliance with different regulations which is needed for any company. Although there are issues with device and network integration and the supporting architecture, the study demonstrates the system's capability of decreasing hospital attendance, increasing medication compliance and quality of life, thereby opening up possibilities for exploring wearable health technologies in clinical environments in the future. © 2024 IEEE.

关键词： Electronic health record

来源：评论

学校读者我要写书评

暂无评论

Face Image Recognition Using Deep learning Through Autoencoder Pre-Training

Face Image Recognition Using Deep learning Through Autoencod...

引用

2023 IEEE International Conference on Modelling, Simulation and Intelligent Computing, MoSICom 2023

作者： Solomon, Enoch Woubie, Abraham Emiru, Eyael Solomon Virginia State University Department of Computer Science RichmondVA United States Silo AI Helsinki Finland Engineering and Computer Science University of Trento Department of Information Italy

ISBN: (纸本)9798350393415

The primary objective of this work is to present an alternative approach aimed at reducing the dependency on labeled data. Our proposed method involves utilizing autoencoder pre-training within a face recognition task with two step processes. Initially, an autoencoder is trained in an unsupervised manner. Subsequently, a deep learning model is trained with initialized parameters from the pre-trained autoencoder. This deep learning training process is conducted in a supervised manner, employing relatively limited labeled training dataset. During evaluation phase, face image embeddings is generated as the output of deep neural network layer. Our training is executed on the CelebA dataset, while evaluation is performed using Labeled Faces in the Wild (LFW) and YouTube Faces (YTF) datasets. Experimental results demonstrate that by initializing the deep neural network with pre-trained autoencoder parameters achieve comparable results to state-of-the-art methods. © 2023 IEEE.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

Soft Prompt Recovers Compressed LLMs, Transferably 41

Soft Prompt Recovers Compressed LLMs, Transferably

引用

41st International Conference on Machine Learning, ICml 2024

作者： Xu, Zhaozhuo Liu, Zirui Chen, Beidi Zhong, Shaochen Tang, Yuxin Wang, Jue Zhou, Kaixiong Hu, Xia Shrivastava, Anshumali Department of Computer Science Stevens Institute of Technology United States Department of Computer Science Rice University United States Department of Electrical and Computer Engineering Carnegie Mellon University United States Together AI United States Department of Electrical and Computer Engineering North Carolina State University United States ThirdAI Corp. United States

Model compression is one of the most popular approaches to improve the accessibility of Large Language Models (LLMs) by reducing their memory footprint. However, the gaining of such efficiency benefits often simultaneously demands extensive engineering efforts and intricate designs to mitigate the performance decline. In this work, we leverage (Soft) Prompt Tuning in its most vanilla form and discover such conventionally learned soft prompts can recover the performance of compressed LLMs. More surprisingly, we observe such recovery effect to be transferable among different tasks and models (albeit natural tokenizer and dimensionality limitations), resulting in further overhead reduction and yet, subverting the common belief that learned soft prompts are task-specific. Our work is fully orthogonal and compatible with model compression frameworks such as pruning and quantization, where we enable up to 8× compressed LLM (with a joint 4-bit quantization and 50% weight pruning compression) to match its uncompressed counterparts on popular benchmarks. We note that we are the first to reveal vanilla Parameter-Efficient Fine-Tuning (PEFT) techniques have the potential to be utilized under a compression recovery context, opening a new line of opportunities for model accessibility advancement while freeing our fellow researchers from the previously present engineering burdens and constraints. The code is available at https://***/zirui-ray-liu/compress-thenprompt. Copyright 2024 by the author(s)

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：