检索结果-内蒙古大学图书馆

arXiv 2022年

作者： Zhao, Zhilin Cao, Longbing Wang, Chang-Dong The Data Science Lab School of Computing and DataX Research Centre Macquarie University SydneyNSW2109 Australia The School of Computer Science and Engineering Sun Yat-sen University Guangzhou China Guangdong Province Key Laboratory of Computational Science Guangzhou China Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education China

The integrity of training data, even when annotated by experts, is far from guaranteed, especially for non-IID datasets comprising both in- and out-of-distribution samples. In an ideal scenario, the majority of samples would be in-distribution, while samples that deviate semantically would be identified as out-of-distribution and excluded during the annotation process. However, experts may erroneously classify these out-of-distribution samples as in-distribution, assigning them labels that are inherently unreliable. This mixture of unreliable labels and varied data types makes the task of learning robust neural networks notably challenging. We observe that both in- and out-of-distribution samples can almost invariably be ruled out from belonging to certain classes, aside from those corresponding to unreliable ground-truth labels. This opens the possibility of utilizing reliable complementary labels that indicate the classes to which a sample does not belong. Guided by this insight, we introduce a novel approach, termed Gray Learning (GL), which leverages both ground-truth and complementary labels. Crucially, GL adaptively adjusts the loss weights for these two label types based on prediction confidence levels. By grounding our approach in statistical learning theory, we derive bounds for the generalization error, demonstrating that GL achieves tight constraints even in non-IID settings. Extensive experimental evaluations reveal that our method significantly outperforms alternative approaches grounded in robust statistics. Copyright © 2022, The Authors. All rights reserved.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Data-driven approach for identifying the factors related to debt collector performance

引用

Journal of Open Innovation: Technology, Market, and Complexity 2024年第4期10卷

作者： Sivamayilvelan, Keerthana Rajasekar, Elakkiya Balachandran, Santhi Kotecha, Ketan Vairavasundaram, Subramaniyaswamy School of Computing SASTRA Deemed University Thanjavur 613401 India xPERT Research Group Department of Computer Science Birla Institute of Technology & Science Pilani - Dubai Campus Dubai 345055 United Arab Emirates School of Arts Sciences Humanities & Education (SASHE) SASTRA Deemed University Thanjavur 613401 India Symbiosis Centre for Applied Artificial Intelligence Symbiosis Institute of Technology Pune India School of Computer Science and Engineering Vellore Institute of Technology Vellore 632014 India

The company's success and growth heavily rely on its workforce's performance, yet the evaluation of employees has been only partially and inconclusively executed so far. The primary goal of this research is to build an open innovation framework for analyzing the performance of the debt collector. We have developed the Reinforcement Learning based Continual Learning (RLC) approach for evaluating the performance by analyzing the metrics such as visit patterns and collection percentage. We have used the private debt collection dataset to assess the debt collector's performance. We formulated hypotheses derived from insights gained during exploratory data analysis and subsequently validated them through statistical testing. Whether there are noticeable distinctions among debt collectors in terms of visitation frequency, collection rates, and collection modes. This proposed open innovation framework for analyzing the debt collector performance provides significant variation in terms of collection rate. The proposed EDQN-CL achieved a 13.56 % higher classification rate than the existing algorithm for categorizing the debt collector performance. © 2024 The Authors

关键词： And machine learning Continual learning Debt collection Debt collector performance Deep reinforcement learning Statistical analysis

来源：评论

学校读者我要写书评

暂无评论

A Machine Learning Model for Augmenting the Media Accessibility for the Disabled People

A Machine Learning Model for Augmenting the Media Accessibil...

引用

Communication, Security and Artificial Intelligence (ICCSAI), International Conference on

作者： Hari Krishnan Andi S. Senbagam Dilip Kumar Sharma Mayura Shelke N K Senthil Kumar Ramesh Chandra Poonia Centre for Postgraduate Studies Asia Metropolitan University Malaysia Department of Computer Science and Engineering Dhanalakshmi Srinivasan College of Engineering and Technology Mamallapuram Tamil Nandu India Department of Mathematics Jaypee University of Engineering and Technology Guna Madhya Pradesh India Department of Artificial Intelligence and Data Science AISSMS IOIT Pune India Department of Computer Science and Engineering School of Computing Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology (Deemed to be University) Chennai Tamil Nadu India Department of Computer Science CHRIST (Deemed to be University) Delhi-NCR Ghaziabad Uttar Pradesh India

In an era characterized by the proliferation of digital media, the need to efficiently use multimedia content has become paramount. This article discusses an innovative technique called “Fast Captioning (FC)” to improve media accessibility, especially for people with disabilities and others with time restrictions. Modern Machine Learning (ML) algorithms are incorporated into the framework, which speeds up video consumption while maintaining content coherence. The procedure includes extracting complex features like Word2Vec embeddings, part-of-speech tags, named entities, and syntactic relationships. Using annotated data, a ML model is trained to forecast semantic similarity scores between words and frames. The predicted scores seamlessly integrate into equations that calculate similarity, thus enhancing content comprehension. Through this all-encompassing approach, the article offers a comprehensive solution that balances the requirements of contemporary media with the accessibility requirements of people with disabilities, producing a more inclusive digital environment. Machine Learning-based Media Augmentation (ML-MA) has achieved the highest accuracy of 96%, and the captioning is accurate.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Investigating Multi-Feature Selection and Ensembling for Audio Classification

arXiv

引用

arXiv 2022年

作者： Turab, Muhammad Kumar, Teerath Bendechache, Malika Saber, Takfarinas Mehran University of Engineering and Technology Jamshoro Pakistan ADAPT – Science Foundation Ireland Research Centre Ireland CRT AI School of Computing Dublin City University Dublin Ireland Lero – the Irish Software Research Centre Ireland School of Computer Science National University of Ireland Galway Ireland

Deep Learning (DL) algorithms have shown impressive performance in diverse domains. Among them, audio has attracted many researchers over the last couple of decades due to some interesting patterns–particularly in classification of audio data. For better performance of audio classification, feature selection and combination play a key role as they have the potential to make or break the performance of any DL model. To investigate this role, we conduct an extensive evaluation of the performance of several cutting-edge DL models (i.e., Convolutional Neural Network, EfficientNet, MobileNet, Supper Vector Machine and Multi-Perceptron) with various state-of-the-art audio features (i.e., Mel Spectrogram, Mel Frequency Cepstral Coefficients, and Zero Crossing Rate) either independently or as a combination (i.e., through ensembling) on three different datasets (i.e., Free Spoken Digits Dataset, Audio Urdu Digits Dataset, and Audio Gujarati Digits Dataset). Overall, results suggest feature selection depends on both the dataset and the model. However, feature combinations should be restricted to the only features that already achieve good performances when used individually (i.e., mostly Mel Spectrogram, Mel Frequency Cepstral Coefficients). Such feature combination/ensembling enabled us to outperform the previous state-of-the-art results irrespective of our choice of DL model. © 2022, CC BY.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

Ledlog: Lightweight Explainable Real-Time Dual-Model Log Anomaly Detection System

SSRN

引用

SSRN 2024年

作者： Tan, Zhuoran Wang, Qiyuan Anagnostopoulos, Christos Marnerides, Angelos K. Singer, Jeremy Parambath, Shameem P. Temple, Sam School of Computing Science University of Glasgow 18 Lilybank Gardens GlasgowG12 8RZ United Kingdom Dept. of Electrical & Computer Engineering KIOS R&I Centre of Excellence University of Cyprus Nicosia Cyprus JUMPSEC Ltd London United Kingdom JUMPSEC Ltd United Kingdom

Log anomaly detection has been a significant asset in cybersecurity, as it can help detect threats at an early stage. Current detectionsystems lack interpretability due to the challenges posed by high volume and data dimensionality, resulting in weakened featureengineering and questioned detection performance. In parallel, the requirement for real-time detection is challenging, given thediverse deployment constraints present in large-scale setups. Equally important is the lack of feasible human-in-the-loop feedbackto support continual learning. To handle these challenges, we propose LEDLog, a system based on an adjustable dual-model detection logic tailored to log characteristics and anomaly *** consists of two binary classification models, employing Multi-layer Perceptron (MLP) and Graph Convolutional Networks (GCNs) tolearn the graph representation of event logs. In order to comprehensively integrate diverse logs, LEDLog incorporates log transformation and pre-processing for unstructured logs. The real-time composition of our schemerelies heavily on feature importance analysis through data dimensionality reduction achieved via appropriate data filtering. Learningmodels are updated through the integration of elastic weight consolidation (EWC) based continual learning. To the best of our knowledge, our mechanism is the first to demonstrate how MachineLearning Operations (MLOPs) principles adapt to local deployment of a log detection system, making it suitable for scenarios requiringextreme privacy. Our experimental evaluation showcases relatively more efficient detection and processing abilities compared toprevious State-of-The-Art (SOTA) works. Thus, LEDLog demonstrates desired properties for operators in the context of steady updates,providing substantial benefits in detection performance. © 2024, The Authors. All rights reserved.

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

Techno-Economic and Emission Analysis of Biomass Integrated Hybrid Energy System (Hes) for an Automotive Condition Monitoring centre

SSRN

引用

SSRN 2024年

作者： Idris, Jamiu Morenikeji Abubakar, Yakubu Khartum Nasir, Abdulkarim Zumami, Ahmad Muhammad Bongfa, Binfa Adelodun, Dauda Jide School of Humanities Social Science Law Centre for Energy Petroleum and Mineral Law and Policy University of Dundee United Kingdom Department of Mechanical Engineering The Federal Polytechnic P.M.B 1037 Kogi State Idah Nigeria Department of Mechanical Engineering Federal University of Technology Minna Nigeria Department of Computer Science The Federal Polytechnic P.M.B 1037 Kogi State Idah Nigeria Department of Advanced Material Engineering School of Computing Engineering and Build Environment Edinburgh Napier University United Kingdom

Nigeria is plague by a lack of electricity, despite her plethora of energy resources. Biogas power plants have been neglected in Nigeria's energy mix. A hybrid energy system (HES) for an automotive condition monitoring centre (ACMC) in Minna, Nigeria, was the focus of techno-economic and emission analysis in this study. After assessing the availability of ACMC load, biomass, and solar energy resources, a simulation software called Hybrid Optimization of Multiple Energy Resources (HOMER) simulates three different system designs: biogas generator-solar PV-battery, solar PV-diesel generator-battery, and diesel generator only. The analysis shown that the ACMC location can yield at least 11 tons of biomass daily. The solar PV-battery-biogas generator design resulted in a Levelized cost of energy (LCOE) of $0.4238 per kWh, a minimal net present cost (NPC) of around $500,000, and can saved the atmosphere around 154,235.78 kg of CO2 per year when compared to running on 100% diesel generator. Additionally, sensitivity analysis showed that the LCOE decreased with a decreasing nominal discount rate and an increasing system life. Integrating biogas generators into Nigeria's energy mix can assist in mitigating power-related issues and can cause a reduction in greenhouse gas (GHG) emissions. © 2024, The Authors. All rights reserved.

关键词： Solar energy

来源：评论

学校读者我要写书评

暂无评论

A Multi-Dimensional Assessment Model and Its Application in E-learning Courses of computer science 20

A Multi-Dimensional Assessment Model and Its Application in ...

引用

21st Annual Conference of the Special Interest Group in Information Technology Education, SIGITE 2020

作者： Luo, Jiwen Lu, Feng Wang, Tao Key Lab. of Parallel and Distributed Computing College of Computer National University of Defense Technology Changsha China National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China

ISBN: (纸本)9781450370455

computer science is a practical discipline. It is always a great challenge to evaluate students' computer practice using computer-aided means for large scale students. We always need to address problems such as suspected plagiarism and deviation of the overall difficulty factor. In this paper, a multi-dimensional assessment model is designed for CS courses based on the detailed practice processing data in an E-learning system. The model comprehensively evaluates the students' learning process and results in three aspects of correctness, originality, and quality detection. Besides, the teacher can easily participate in the assessment according to their needs. The correctness is an essential requirement, and the originality is based on the clustering results of students' behaviors after clone detection to curb homework plagiarism. SonarQube is used to detect code quality and put forward higher requirements for codes. Manual participation intelligence has improved the flexibility and applicability of the model to a certain extent. We applied this model on the EduCoder online education platform and carried out a comprehensive analysis of 485 students in the parallel Programming Principles and Practice Class of Huazhong University of science and Technology. Experiment results confirm the distinction, rationality, and fairness of the model in assessing student performance. It not only gives students a credible, comprehensive score in large-scale online practical programming courses but also gives teachers and students corresponding suggestions based on the evaluation results. Furthermore, the model can be extended to other online education platforms. © 2020 ACM.

关键词： Students

来源：评论

学校读者我要写书评

暂无评论

Optimizing Fire Safety: Reducing False Alarms Using Advanced Machine Learning Techniques

arXiv

引用

arXiv 2025年

作者： Jamal, Muhammad Hassan Alazeb, Abdulwahab Bakhsh, Shahid Allah Boulila, Wadii Shah, Syed Aziz Khattak, Aizaz Ahmad Khan, Muhammad Shahbaz Department of Computer Sciences Quaid-i-Azam University Islamabad45320 Pakistan Department of Computer Science College of Computer Science and Information Systems Najran University Najran61441 Saudi Arabia Department of Cyber Security National University of Science & Technology Karachi75350 Pakistan College of Computer and Information Sciences Prince Sultan University Riyadh11586 Saudi Arabia Centre for Intelligent Healthcare Coventry University CoventryCVl 5FB United Kingdom School of Computing Engineering and the Built Environment Edinburgh Napier University EdinburghEHlO 5DT United Kingdom

Fire safety practices are important to reduce the extent of destruction caused by fire. While smoke alarms help save lives, firefighters struggle with the increasing number of false alarms. This paper presents a precise and efficient Weighted ensemble model for decreasing false alarms. It estimates the density, computes weights according to the high and low-density regions, forwards the high region weights to KNN and low region weights to XGBoost and combines the predictions. The proposed model is effective at reducing response time, increasing fire safety, and minimizing the damage that fires cause. A specifically designed dataset for smoke detection is utilized to test the proposed model. In addition, a variety of ML models, such as Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), Nai:ve Bayes (NB), K-Nearest Neighbour (KNN), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost), Adaptive Boosting (ADAB), have also been utilized. To maximize the use of the smoke detection dataset, all the algorithms utilize the SMOTE re-sampling technique. After evaluating the assessment criteria, this paper presents a concise summary of the comprehensive findings obtained by comparing the outcomes of all models. © 2025, CC BY-NC-ND.

关键词： Smoke detectors

来源：评论

学校读者我要写书评

暂无评论

Evaluating Security of MQTT Protocol in Internet of Things

Evaluating Security of MQTT Protocol in Internet of Things

引用

Canadian Conference on Electrical and computer engineering (CCECE)

作者： Ayman Al-Ani Wong Kang Shen Ahmed K. Al-Ani Shams A. Laghari Omar E. Elejla Cybersecurity Research Group Faculty of Computing and Informatics Universiti Malaysia Sabah Jalan UMS Kota Kinabalu Sabah Malaysia School of Computing and Data Science Xiamen University Malaysia Sepang Selangor Malaysia School of Engineering Technology and Applied Science Centennial College Toronto Ontario Canada National Advanced IPv6 Centre (NAv6) Universiti Sains Malaysia (USM) Gelugor Penang Malaysia Department of Computer Science Al-Aqsa University Gaza Palestine

The Internet of Things (IoT) has revolutionized the way people interact, communicate, and perform daily activities in various domains ranging from households to industries and cities. MQTT is one of the commonly adopted protocols for implementing IoT. However, IoT systems that are connected through MQTT are susceptible to security breaches as MQTT was not originally designed with security as a priority. The credentials and messages transmitted in plaintext by default, thereby compromising data confidentiality and integrity. This study presents a comprehensive analysis of the MQTT protocol, including experimentation on an MQTT system using various cryptographic implementations, such as AES-CBC, RSA, and ECC AES Hybrid Scheme, to assess the processing time and message size. The findings indicate that payload encryption increases processing time and message bytes. Among the cryptographic implementations, RSA incurs the highest processing time, followed by ECC AES Hybrid Scheme and AES- CBC. Furthermore, the study demonstrates the effectiveness of attack prevention between standard MQTT and secured MQTT implementations by simulating various IoT attacks, such as black-box penetration attack, identity spoofing, DoS attack, and MITM attack. The results and subsequent discussion provide insights that answer the research question, revealing the cryptographic algorithms that result in the most overhead on the standard MQTT implementation and their capacity to resist common attacks.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Advantages of Synchronizing Vehicles Intersection Access

引用

Transportation Research Procedia 2024年 78卷 475-482页

作者： Radha Reddy Luis Almeida Pedro Santos Eduardo Tovar CISTER Research Centre Rua Alfredo Allen 535 4200-135 Porto Portugal Instituto Superior de Engenharia do Porto (ISEP) R. Dr. António Bernardino de Almeida 431 4200-135 Porto Portugal Faculty of Engineering of University of Porto (FEUP) Rua Dr. Roberto Frias 4200-465 Porto Portugal Dept. of Computer Science and Engineering Amrita School of Computing Coimbatore Amrita Vishwa Vidyapeetham India

An intersection area is a common space that must be efficiently shared over time between vehicles with conflicting trajectories from several road lanes. Numerous traffic light control (TLC) strategies have been presented to address this. The main objective is periodically switching intersection access between road lanes by permitting vehicles sequentially, parallelly, or synchronously. This work addresses which intersection access type is more efficient for throughput, travel delays, fuel consumption, and associated tailpipe emissions. We compare the performance of five state-of-the-art TLC approaches in two types of road networks. Among the five TLC approaches, three are sequential, one is parallel, and the other is synchronous. In the first road network, all four intersections are homogeneous with four legs. The second network is heterogeneous, with two intersections with four legs and the other two with three legs, i.e., T-intersections. We also consider two maximum speeds (30 and 50km/h) representing low-speed urban settings. SUMO simulation results suggest all access types have similar throughput, with a minor advantage of synchronous access (up to 3%). However, compared to the other best approaches, synchronous intersection access significantly reduces travel time loss (up to 130.5%) and fuel consumption (up to 37.2%).

关键词： Vehicles intersection access sequential parallel synchronous throughput travel time loss and fuel efficiency

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：