检索结果-内蒙古大学图书馆

IEEE Industry Applications Society Annual Meeting (IAS)

作者： Ebrahim Balouji Karl Bäckstrüm Tomas McKelvey Dept. of Electrical Engineering Chalmers University of Technology Gothenburg Sweden Dept. of Computer Science and Engineering Chalmers University of Technology Gothenburg Sweden

This research work introduces Deep Deterministic Policy Gradient (DDPG), a type of Reinforcement Learning (RL), for grid modeling, estimating voltage and phase angle, and control method for grid-forming inverters. The aim is to develop a grid-forming inverter that sets the voltage level and frequency of the grid and mitigates voltage dips originating from faults and frequency deviation fluctuations. Unlike conventional methods for estimating setpoints for the controller loops, we do not need several chains of estimation tools such as Fast Fourier Transform (FFT), Synchronous Reference Frame (SRF), or lowpass filters. With the DDPG, we also optimize the phase lock-loop (PLL) and accurately deliver the angle for the actuation part of the inverter to generate the given reference signal. The developed method does not need exhaustive tuning of parameters such as coefficients of PID controllers and lowpass filters. We observe that the proposed method has a faster response time than the PID-based control unit (15ms compared to 50ms) for the grid-forming inverter in the case of compensating voltage dips. We also observed that the DDPG-based grid-forming inverter is more efficient in compensating continuous voltage variations and frequency deviations than a trivial PID-based version.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Exploring the Impact of Conditional Deep Convolutional Generative Adversarial Networks in Brain Tumor Image Classification: A Novel Approach

Exploring the Impact of Conditional Deep Convolutional Gener...

引用

Pervasive Computational Technologies (ICPCT), International Conference on

作者： Deepak Upadhyay Abhay Upadhyay Kuj Bihari Sharma Shiv Ashish Dhondiyal Nookala Venu Dept. of Computer Science and Engineering Graphic Era Hill University Dehradun India Dept. of Computer Science and Engineering Graphic Era Deemed to be University Dehradun India Centre for Internet of Things Madhav Institute of Technology & Science Deemed University Gwalior India

ISBN: (数字)9798331508685

ISBN: (纸本)9798331519476

In this paper, we explore the impact of conditional Deep Convolutional Generative Adversarial Networks (cDCGANs) with brain tumor image classification and introduce a new way to improve diagnosis accuracy. Shortage of Data/Data Imbalance: In Medical Imaging datasets, it helps in surging the data scarcity and class imbalance with mimicking generation images using cDCGAN. The detailed investigation demonstrates that synthetic data effectively improves classifier performance by acting as a form of augmentation, providing additional training examples for classifiers and serving as ground truth. From the computational efficiency analysis, we see that there is also scalability in training time with respect to dataset size. Consequently, this study presents cDCGANs as an important asset with which to improve diagnostic accuracy through brain tumor categorization in clinical contexts.

关键词： Training Accuracy Scalability Brain tumors Medical services Sensitivity and specificity Generative adversarial networks Robustness Image classification Synthetic data

来源：评论

学校读者我要写书评

暂无评论

Light-Weight Gated Recurrent Unit for Electroencephalogram Emotion Classification

Light-Weight Gated Recurrent Unit for Electroencephalogram E...

引用

International Electrical Engineering Congress (iEECON)

作者： Suchada Sitjongsataporn Wenzhang Du Pipat Sakarin Dept. of Electronic Engineering Mahanakorn Institute of Innovation (MII) Faculty of Engineering and Technology Mahanakorn University of Technology Bangkok Thailand Dept. of Computer Engineering Mahanakorn University of Technology International College (MUTIC) Faculty of Engineering and Technology Mahanakorn University of Technology

ISBN: (数字)9798331543952

ISBN: (纸本)9798331543969

This experiment investigates the impact of reducing GRU units and dense layer neurons in a lightweight GRU architecture (LW-GRU-RU) compared to a baseline GRU model for electroencephalogram (EEG) emotion classification. A baseline GRU model is used as a reference, with an optimized GRU variant utilizing feature selection through a Random Forest-based algorithm. Experiments are conducted on a labeled emotion dataset, comparing accuracy and training efficiency across five trials. Results highlight the trade-off between model complexity, accuracy, and computational efficiency, providing insights for practical applications. Both models are evaluated on an emotion dataset for accuracy and training efficiency. The lightweight model achieves a competitive accuracy of 97.486% while reducing the average training time to 0.19 seconds per epoch, showcasing its potential for efficient real-world applications.

关键词： Training Hands Accuracy Computational modeling Neurons computer architecture Brain modeling Feature extraction Electroencephalography Computational efficiency

来源：评论

学校读者我要写书评

暂无评论

A Novel Technique to Parameterize Congestion Control in 6TiSCH IIoT Networks

A Novel Technique to Parameterize Congestion Control in 6TiS...

引用

IEEE Conference on Global Communications (GLOBECOM)

作者： Kushal Chakraborty Aritra Kumar Dutta Mohammad Avesh Hussain Syed Raafay Mohiuddin Nikumani Choudhury Rakesh Matam Mithun Mukherjee Dept. of Computer Science & Information Systems Birla Institute of Technology & Science Hyderabad India Dept. of Computer Science & Engineering Indian Institute of Information Technology Guwahati India Nanjing University of Information Science and Technology Nanjing China

The Industrial Internet of Things (IIoT) refers to the use of interconnected smart devices, sensors, and other technologies to create a network of intelligent systems that can monitor and manage industrial processes. 6TiSCH (IPv6 over the Time Slotted Channel Hopping mode of IEEE 802.15.4e) as an enabling technology facilitates low-power and low-latency communication between IoT devices in industrial environments. The Routing Protocol for Low power and lossy networks (RPL), which is used as the de-facto routing protocol for 6TiSCH networks is observed to suffer from several limitations, especially during congestion in the network. Therefore, there is an immediate need for some modifications to the RPL to deal with this problem. Under traffic load which keeps on changing continuously at different instants of time, the proposed mechanism aims at finding the appropriate parent for a node that can forward the packet to the destination through the least congested path with minimal packet loss. This facilitates congestion management under dynamic traffic loads. For this, a new metric for routing using the concept of exponential weighting has been proposed, which takes the number of packets present in the queue of the node into account when choosing the parent at a particular instance of time. Additionally, the paper proposes a parent selection and swapping mechanism for congested networks. Performance evaluations are carried out in order to validate the proposed work. The results show an improvement in the performance of RPL under heavy and dynamic traffic loads.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Beyond Sequence: Advanced Homology Detection for Cancer Protein using MS_HITAMPE

引用

Procedia computer Science 2025年 258卷 213-222页

作者： S. Rajarajeswari S. Kannan Research Scholar MKU24FFOS11386 Madurai Kamaraj University (MKU) Madurai Professor Dept. of Computer Applications School of Information Technology MKU Madurai

A key challenge in bioinformatics continues to be the detection of remote homologies between cancer proteins, which has important implications for the understanding of disease mechanisms and development of targeted therapies. Using MS_HITAMPE (Multi-Scale Homology Identification Technique with Alignment and Matching for Protein Evolution), this study identifies distant evolutionary relationships among cancer-related proteins that surpass traditional sequence-based methods. In this research, MS_HITAMPE was developed, which combines local and global sequence features to detect subtle evolutionary signals that conventional approaches miss. Detection of remote homologs of cancer proteins has been made more sensitive by MS_HITAMPE, which detects remote homologs with sequence similarities as low as 20%, which is a significant improvement over existing method. The results of a rigorous performance evaluation demonstrate MS_HITAMPE’s enhanced accuracy (an improvement of 15%), decreased time complexity (30% faster), and greater search similarity ratio (25%). Analysing 10,000 cancer-associated proteins from the Protein Data Bank (PDB) revealed 500 previously unknown homologous relationships, providing new insight into oncogenic protein evolution. Using MS_HITAMPE to predict the functions of uncharacterized cancer proteins, with 85% validation accuracy. In this work, researchers are given a powerful tool for analysing the complex landscape of cancer proteomics, representing a significant advance in computational biology. As a result of the discovery of hidden evolutionary connections, MS_HITAMPE can be used to uncover new hypotheses about cancer biology and potential therapeutic targets. In addition to cancer research, the method may enhance knowledge of protein evolution and function across a variety of biological fields.

关键词： Cancer Protein Homology Protein function prediction Remote homology detection Sequence alignment Multi-Scale Homology Identification Technique

来源：评论

学校读者我要写书评

暂无评论

Proposed Model for Prediction of Stock Market Price of Netflix

Proposed Model for Prediction of Stock Market Price of Netfl...

引用

Artificial Intelligence and Applications (ICAIA), International Conference on, technology Conference (ATCON-1), Alliance

作者： Pratap Singh Patwal Anoop Kumar Srivastava Dept. of Computer Science & Engineering Laxmi Devi Institute of Engineering & Technology Alwar Rajasthan India Dept. of Computer Science & Engineering ACED Alliance University Bengaluru Karnataka India

Accurate prediction of stock market price is highly challenging. This paper presents a proposed model for prediction of stock market price of Netflix. We have considered a five–year data set (April, 2017 – April, 2022) of Netflix. An Exploratory Data Analysis (EDA) of Netflix’s stock price data for predicting its stock market prices using time series is done. The implementation of the model is done using Python language. We imported five-years data and applied several techniques: importing libraries, calculating stock return, line plot, plot all, plot return year wise, plot histogram, plot kernel density, plot box plot, differencing method, resample daily to monthly data etc. EDA proved that using time series technique achieved better results in prediction of stock price and visualizing.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Deep Learning based Bangladeshi Vehicle Classification using Fine-Tuned Multi-class Vehicle Image Network (MVINet) Model

A Deep Learning based Bangladeshi Vehicle Classification usi...

引用

Next-Generation Computing, IoT and Machine Learning (NCIM), International Conference on

作者： Tanvir Ahmed Farzana Sharmin Mou Abu Anas Ibn. Samad Dept. of Computer Science and Engineering Dhaka University of Engineering & Technology Dhaka Gazipur Bangladesh Dept. of Computer Science University of Birmingham Birmingham UK Computer Vision Engineering Lab Sigmind.ai Limited Dhaka Bangladesh

The classification of vehicles is an important task in many applications, including traffic management, surveillance, and law enforcement. In Bangladesh, vehicle classification has become increasingly challenging due to the rapid growth in the number of vehicles on the roads. In this study, a novel deep learning-based Bangladeshi vehicle classification model using fine-tuned Multi-class Vehicle Image Network (MVINet) based on DenseNet201 with four added layers has been proposed. Our approach uses convolutional neural networks (CNNs) to extract features from images of vehicles. To make our dataset, we employ a YOLO (You Only Look Once) model for initial vehicle detection, followed by manual filtering and augmentation of the dataset to improve classification accuracy. The proposed method demonstrates superior performance compared to manual data collection methods. We evaluated our approach using standard metrics such as accuracy, precision, recall, and F1 score. Our model achieved an accuracy of 97.06%, precision of 97.17%, recall of 97.24%, and F1-score of 97.12% indicating high performance in classifying vehicles. We also compared our approach with state-of-the-art techniques and found that the proposed deep learning-based approach outperformed them significantly. The proposed approach has the potential to improve traffic management and law enforcement in Bangladesh. It can be used to monitor traffic flow, detect traffic violations, and identify stolen vehicles. Overall, this study demonstrates the effectiveness of deep learning in Bangladeshi vehicle detection and classification and highlights the importance of using advanced technologies to address the challenges of a rapidly changing transportation landscape.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Feature Extraction Based Acoustic Signal Detection in a Cost Effective Φ-OTDR System

Feature Extraction Based Acoustic Signal Detection in a Cost...

引用

International Conference on Microwave, Optical and Communication Engineering (ICMOCE)

作者： Shruti Verma Jimson Mathew Sumanta Gupta Dept. of Electrical Engineering Indian Institute of Technology Patna India Dept. of Computer Science and Engineering Indian Institute of Technology Patna India

The paper presents an experimental demonstration of acoustic signal detection in a phase-optical time domain reflectometer (Φ-OTDR) system based on feature extraction. A commercially available off-the-shelf distributed feedback (DFB) laser is used as a light source with a minimum of 25kHz of linewidth. Also, a low sampling rate (60MS/s) data acquisition (DAQ) system is used to collect the backscattered traces. Using the digital signal processing (DSP) technique, the noisy signal from the Φ-OTDR system is processed to extract the different features of the signal. The location of the acoustic signal source is identified based on variance along the fiber length. Besides, power spectral density at that perturbed location is estimated to determine the acoustic signals frequencies. Using this low cost experimental set-up, the location of the acoustic signal source having 100Hz, 125Hz, and 150Hz frequencies were successfully detected over a 1 km of fiber length.

关键词：

来源：评论

学校读者我要写书评

暂无评论

AI-based botnet attack classification and detection in IoT devices

AI-based botnet attack classification and detection in IoT d...

引用

2022 IEEE International Conference on Machine Learning and Applied Network Technologies, ICMLANT 2022

作者： Puri, Vikram Kataria, Aman Solanki, Vijender Kumar Rani, Sita Duy Tan University Center of Visualization and Simulation Da Nang Viet Nam CSIR-CSIO Chandigarh India CMR Institute of Technology Dept. of Computer Science and Engineering TS Hyderabad India Guru Nanak Dev Engineering College Dept. of Computer Science and Engineering Punjab Ludhiana India

ISBN: (纸本)9781665487016

End-user Internet of Things (IoT) devices, including security cameras, smart appliances, home monitors, and thermostats, are becoming more prevalent in households. Additionally, the proliferation of devices facilitates the propagation of security concerns like DoS and spoofing. However, it is difficult for conventional rule-based security systems to recognize IoT assaults due to the development of heterogenous devices in the IoT ecosystem. Artificial Intelligence (AI) techniques can be a solution which enables the creation of an effective security model based on actual data from each device. In this work, IoT botnets are detected and classified using machine learning (ML) and deep learning (DL) based algorithms. Six ML models and three DL models are used to assess the system's performance. The best-performing model is also implemented as an API. © 2022 IEEE.

关键词： Internet of things

来源：评论

学校读者我要写书评

暂无评论

Traffic Light Control Using Reinforcement Learning 2

Traffic Light Control Using Reinforcement Learning

引用

2nd IEEE International Conference on Integrated Circuits and Communication Systems, ICICACS 2024

作者： Masfequier Rahman Swapno, S.M. Nuruzzaman Nobel, S.M. Ramachandra, A.C. Islam, Md Babul Haque, Rezaul Rahman, Mohammad Mominur Bangladesh University of Business and Technology Dept of CSE Dhaka Bangladesh Nitte Meenakshi Institute of Technology Department of Electronics and Communication Engineering Bengaluru India UNICAL Dept. of Computer Modeling Electronic and System Engineering Rende Italy East West University Dept of CSE Dhaka1212 Bangladesh Hamad Bin Khalifa University College of Science and Engineering Qatar

ISBN: (纸本)9798350317558

Nowadays, one of the biggest issues in urban areas is traffic. This problem wastes important time and contributes to air and sound pollution. This affects people's general quality of life in addition to posing health dangers. Our study attempts to mitigate these problems, effectively cutting down on wait times and delays. Our Reinforcement Learning method creates intelligent agents that can adjust traffic lights at crossings instantly. Our objective is to minimize delays, reduce congestion, reduce travel times, improve safety, and improve traffic flow. We implemented the Deep Q Learning algorithm which activities yield the greatest benefits under various traffic scenarios. Our model can the sequence time since the Green signal (GS) lasts 10 seconds and the Red signal (RS) lasts 5 seconds. The waiting period is shortened by 50% as a result. This study suggests reinforcement learning may improve traffic signal controller synchronization and urban traffic congestion. This novel method may improve transport efficiency and sustainability. © 2024 IEEE.

关键词： Action Deep Q Learning GS Model Q Value Rewards RS State Traffic signal vehicles Waiting Time

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：