检索结果-内蒙古大学图书馆

19th International Conference on Advanced Data Mining and Applications, ADMA 2023

作者： Tekawade, Atharva Banerjee, Suman Department of Computer Science and Engineering Indian Institute of Technology Jammu Jammu181221 India

ISBN: (纸本)9783031466601

Given a Social Network how to select a small number of influential users to maximize the influence in the network has been studied extensively in the past two decades and formally referred to as the Influence Maximization Problem. Among most of the existing studies, it has been implicitly assumed that there exists a single probability value that represents the influence probability between the users. However, in reality, the influence probability between any two users is dependent on the context (formally referred to as tag e.g.;a sportsman can influence his friends related to any news related to sports with high probability). In this paper, we bridge the gap by studying the Tag-Based Influence Maximization Problem. In this problem, we are given with a social network where each edge is marked with one probability value for every tag and the goal here is to select k influential users and r influential tags to maximize the influence in the network. First, we define a tag-based influence function and show that this function is bi-submodular. We use the orthent-wise maximization procedure of bi-submodular function which gives a constant factor approximation guarantee. Subsequently, we propose a number of efficient pruning techniques that reduces the computational time significantly. We perform an extensive number of experiments with real-world datasets to show the effectiveness and efficiency of the proposed solution approaches. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

关键词： Method of moments

来源：评论

学校读者我要写书评

暂无评论

Analyzing Hierarchical Relationships and Quality of Embedding in Latent Space

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2025年第4期6卷 843-858页

作者： Chatterjee, Ankita Mukherjee, Jayanta Das, Partha Pratim Indian Institute of Technology Kharagpur Department of Computer Science and Engineering Kharagpur721302 India

Existing learning models partition the generated representations using hyperplanes which form well defined groups of similar embeddings that is uniquely mapped to a particular class. However, in practical applications, the embedding space does not form distinct boundaries to segregate the class representations. There exists interaction among similar classes which cannot be visually determined in high-dimensional space. Moreover, the structure of the latent space remains obscure. As learned representations are frequently reused to reduce the inference time, it is important to analyse how semantically related classes interact among themselves in the latent space. Therefore, we propose a boundary estimation algorithm that minimises the inclusion of other classes in the embedding space to form groups of similar representations and compare the quality of these class embeddings for various models in an already encoded space. These groups are overlapping to denote ambiguous embeddings that cannot be mapped to a particular class with high confidence. The algorithm determines which representations to be included or discarded to form well defined regions, separating discriminating, ambiguous and rejected embeddings to depict a particular class. Later, we construct relation trees to evaluate the hierarchical relationships formed among the classes, and compare it with the WordNet ontology using phylogenetic tree comparison methods. © 2020 IEEE.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

A novel binary quantizer for variational autoencoder-based image compressor

引用

International Journal of computers and Applications 2024年第8期46卷 604-620页

作者： Thulasidharan, Pillai Praveen Nath, Keshab Department of Computer Science and Engineering Indian Institute of Information Technology Kottayam India

Neural network-based encoder and decoder are one of the emerging techniques for image compression. To improve the compression rate, these models use a special module called the quantizer that improves the entropy of the generated representation. Here, we propose a novel binary quantizer and an auxiliary support module known as the remapper that co-ordinate for better compression and decompression of an image. The proposed quantizer generates the compressed binary representation whereas the remapper aids in minimizing the reconstruction error during decompression. The combined optimization of the model consisting of the encoder, quantizer, remapper, and the decoder as a single entity is not feasible due to the non-differentiable nature of the quantizer. To mitigate this, we propose a step-by-step training process wherein the encoder and decoder are trained first, then keeping the encoder parameters fixed, the remapper is trained along with the decoder to obtain a fully optimized model. The training is performed with different types of images and the results are compared to understand the best training procedures. Experimental results reveal that the proposed model can compress a (Formula presented.) image to a mere 256 bits, thereby achieving a compression ratio of 6144:1;with excellent perceptual reconstruction during decompression. © 2024 Informa UK Limited, trading as Taylor & Francis Group.

关键词： Image compression

来源：评论

学校读者我要写书评

暂无评论

Automatic Creation of Ontologies and Rules from ML Models for Access Control in IoT 15

Automatic Creation of Ontologies and Rules from ML Models fo...

引用

15th International Conference on Computing Communication and Networking Technologies, ICCCNT 2024

作者： Tyagi, Surendra Jinwala, Devesh C. Bhattacharjee, Subhasis Computer Science & Engineering Indian Institute of Technology Jammu India Computer Science & Engineering S. V. National Institute of Technology Surat India

ISBN: (纸本)9798350370249

Ontologies have been increasingly used in designing access control policies in Internet of Things (IoT) systems. Machine Learning (ML) has been extensively used to identify patterns from existing data. Trained ML models can be used on end devices for applying access control for resources and users. However it is unreasonable to deploy ML models into the low-powered and mostly hand-held IoT devices. This paper demonstrates an attempt to exploit the power of ML for generating access control rules. We propose algorithms for generating Ontologies and policy rules from trained ML models. Motivated by their simple design, low or almost no requirement for data preprocessing, no requirement for scaling/normalization of data, simple, intuitive design, and most essentially an inherent ability to optimize, by handling irrelevant features through pruning, we use Decision Trees (DT) for the classification in the proposed approach. For smart home dataset, we show in this paper, automatic creation of Ontologies and policy rules from trained DT models. The resulting Ontologies and Semantic Web Rule Language (SWRL) rules are then deployed into an Ontological framework (Protege) and tested for consistency using a Reasoner. Our approach demonstrates a significant improvement in the efficiency of rule generation and offers a promising direction towards designing automated access control systems. In Protege, we achieve 100% consistency in deployment and testing of the generated access control policies by firing queries. We obtain exactly the same accuracy from Protege queries as we get from ML model. We also propose pruning on DT as a fall-back method for IoT devices with limited computing and memory resources that can handle only limited number of policy rules. The proposed method creates a deployable access control system in a few minutes without any human intervention. Our experiment shows a turnaround time of 5 minutes for the Smart Home dataset. ©2024 IEEE.

关键词： Decision trees

来源：评论

学校读者我要写书评

暂无评论

EPASAD:ellipsoid decision boundary based Process-Aware Stealthy Attack Detector

引用

Cybersecurity 2024年第3期7卷 32-48页

作者： Vikas Maurya Rachit Agarwal Saurabh Kumar Sandeep Shukla Department of Computer Science and Engineering Indian Institute of Technology KanpurkanpurIndia Merkle Science BangaloreIndia

Due to the importance of Critical Infrastructure(Cl)in a nation's economy,they have been lucrative targets for cyber *** critical infrastructures are usually Cyber-Physical Systems such as power grids,water,and sewage treatment facilities,oil and gas pipelines,*** recent times,these systems have suffered from cyber attacks numer-ous *** have been developing cyber security solutions for Cls to avoid lasting *** to standard frameworks,cyber security based on identification,protection,detection,response,and recovery are at the core of these *** of an ongoing attack that escapes standard protection such as firewall,anti-virus,and host/network intrusion detection has gained importance as such attacks eventually affect the physical dynamics of the ***,anomaly detection in physical dynamics proves an effective means to implement *** is one example of anomaly detection in the sensor/actuator data,representing such systems physical *** present EPASAD,which improves the detection technique used in PASAD to detect these micro-stealthy attacks,as our experiments show that PASAD's spherical boundary-based detection fails to *** method EPASAD overcomes this by using Ellipsoid boundaries,thereby tightening the boundaries in various dimen-sions,whereas a spherical boundary treats all dimensions *** validate EPASAD using the dataset produced by the TE-process simulator and the C-town *** results show that EPASAD improves PASAD's average recall by 5.8%and 9.5%for the two datasets,respectively.

关键词： Intrusion detection system Critical infrastructure security Industrial control system Machine learning

来源：评论

学校读者我要写书评

暂无评论

TREAT: Temporal and Relational Attention-Based Tensor Representation Learning for Ethereum Phishing Users

引用

IEEE Transactions on Services Computing 2025年第03期18卷 1838-1851页

作者： Ghosh, Medhasree Halder, Raju Chandra, Joydeep Indian Institute of Technology Patna Deprtment of Computer Science and Engineering India

The Ethereum blockchain platform has witnessed a surge in crypto-cybercrimes, particularly phishing attacks, resulting in significant financial losses. Analyzing the Ethereum transaction network to detect phishing users poses a set of critical challenges, including network sparsity, dynamic network fluctuations, large-scale data and significant class imbalance. Existing literature in this area primarily leverages traditional feature engineering or network representation learning to retrieve crucial information from transaction records to identify suspected users. However, these methods mainly rely on manually handcrafted features or conventional node representation learning while ignoring the inherent network sparsity and dynamic fluctuations. Hence, to alleviate these challenges, this paper introduces a novel tensor-based representation learning framework, TREAT (Temporal and Relational Attention-based Tensor Representation Learning). TREAT models the Ethereum transaction network as a 3-dimensional tensor to preserve structural, transactional, and temporal aspects in a standalone architecture, thereby observing the rich correlation among these. The framework is coupled with a two-way self-attention mechanism alongside a rank-based tensor decomposition to comprehend the underlying evolving transaction interaction patterns while addressing the network sparsity. A Graph Neural Network layer with edge attention elevates the final representation quality, thereby yielding a 3%~4% improvement in F1-Score with respect to the existing baselines. © 2008-2012 IEEE.

关键词： Phishing

来源：评论

学校读者我要写书评

暂无评论

Forensic Analysis and Detection of Spoofing Based Email Attack Using Memory Forensics and Machine Learning 1

引用

18th EAI International Conference on Security and Privacy in Communication Networks, SecureComm 2022

作者： Shukla, Sanjeev Misra, Manoj Varshney, Gaurav Department of Computer Science and Engineering Indian Institute of Technology Roorkee India Department of Computer Science and Engineering Indian Institute of Technology Jammu India

ISBN: (数字)9783031255380

ISBN: (纸本)9783031255373

Emails encounter many types of cyber-attacks and email spoofing is one of the most common and challenging investigation problems. This paper identifies spoofing-based email attacks in an organization by analyzing received and replied emails. The detection works by capturing the email traces via memory forensics. Unlike the traditional approaches of capturing the entire physical memory, we only capture the memory of relevant processes for email header extraction. It significantly reduces the size of the memory dump and makes detection faster. We suggest a novel mechanism called URL extractor, which uses seven novel features from URL to identify the live running email message process by applying ML that traces received emails and captures their header fields for analysis. The authentication header fields of SPF, DKIM, DMARC, and ARC are examined closely to develop a detection algorithm for received emails. Similarly, novel header fields of Reference along with MX record are applied for the detection of replied emails. The MX record is fetched to verify the domain name by sending a forward ns-lookup query to DNS. It also includes an email attack alert mechanism for intimating IT admins of an organization regarding suspected attacks. The results thus obtained show that email detection takes 35 secs (apprx.) to complete with high accuracy and low false positives. © 2023, ICST institute for computer sciences, Social Informatics and Telecommunications engineering.

关键词： Cybersecurity

来源：评论

学校读者我要写书评

暂无评论

A multi-fused convolutional neural network model for fruit image classification

引用

International Journal of Cognitive Computing in engineering 2024年第1期5卷 416-424页

作者： Sinha, Bam Bahadur Dhanalakshmi, R. Computer Science and Engineering National Institute of Technology Sikkim 737139 India Computer Science and Engineering Indian Institute of Information Technology Tiruchirappalli Tamil Nadu 620012 India

Fruit categorization presents a significant challenge due to the diverse range of fruit types and their similarities in color, shape, size, and structure. This challenge is addressed in this research by proposing a multi-fused CNN approach that integrates three state-of-the-art deep learning models, namely EfficientNet-B0, MobileNetV2, and ResNet50V2, into a unified pipeline. The primary objective is to enhance prediction accuracy and mitigate overfitting issues commonly encountered with conventional techniques. Utilizing the Fruit-360 dataset (both small and full versions), the proposed multi-fused CNN model demonstrates superior performance, achieving an impressive accuracy rate of 99.32% and 97.15% in classifying fruit types across 24 and 131 different categories, respectively. This substantial improvement underscores the potential of the proposed approach in reliably detecting fruit types. In future work, efforts will be directed towards optimizing the model for enhanced processing speed, facilitating real-time performance in practical applications, particularly in agricultural automation. Additionally, rigorous experiments on diverse real-world fruit images will be conducted to further validate the robustness of the proposed model. © 2024 The Authors

关键词： Fruits

来源：评论

学校读者我要写书评

暂无评论

Cellular Automaton–Based Sentiment Analysis for Bipolar Classification of Reviews

引用

Complex Systems 2025年第1期34卷 29-57页

作者： Elizabeth, M.J. Chaudhuri, Parimal Pal Hazari, Raju Department of Computer Science and Engineering National Institute of Technology Kerala Calicut India Indian Institute of Technology Kharagpur India

One of the main goals of sentiment analysis is to analyze human perception to continuously adapt to each person’s demands. The information gathered is structured to understand the mood or emotional tone of the reviews and comments. In every aspect of society, there is always an opportunity for opinions and suggestions from the public. So, the quality of the services provided by various resources can considerably increase if we can determine whether those thoughts are positive, negative or neutral. In this paper, we employ cellular automata (CAs) to categorize the reviews as positive or negative. The fundamental advantage of the suggested technique is that there is less need to be concerned about the linguistic characteristics of human languages. The proposed method performs better by leveraging a variety of parameters that are extracted by analyzing cycle length (CL) graphs generated using CL signals produced by elementary CAs (ECAs) without employing any form of pre-trained language model. We evaluated the effectiveness of the proposed cellular automaton–based machine learning (CAML) classifier by assessing certain parameters such as recall, precision, F1 score and accuracy. Our method achieves a maximum accuracy of 92.44% for the sentiment tweet dataset. We compared the proposed technique with all existing machine learning models, as well as recent methodologies employed in the field. The suggested method performs well in predicting the positive and negative classes of sentiments using CAs. © 2025, Complex Systems Publications, Inc. All rights reserved.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Integrated Crowd Counting System Utilizing IoT Sensors, OpenCV and YOLO Models for Accurate People Density Estimation in Real-Time Environments 1

Integrated Crowd Counting System Utilizing IoT Sensors, Open...

引用

1st International Conference on Cognitive, Green and Ubiquitous Computing, IC-CGU 2024

作者： Jayasingh, Suvendra Kumar Naik, Pintu Swain, Satyaprakash Patra, Kumar Janardan Kabat, Manas Ranjan Institute of Management and Information Technology BPUT Department of Computer Science and Engineering Odisha Cuttack India Indian Institute of Technology Department of Computer Science and Engineering Jammu India

ISBN: (纸本)9798350361780

Real-time crowd monitoring plays a pivotal role in effectively managing public spaces and ensuring safety. This study investigates the fusion of IoT devices and the YOLO object detection model to accurately count crowds. IoT devices facilitate the instantaneous collection of data from cameras, while the YOLO model adeptly identifies individuals within recorded video frames. The study rigorously assesses the performance of three YOLO variants: YOLO V5, YOLO V8 and YOLO V8 NAS. Findings reveal that YOLO V8 NAS surpasses YOLO V5 and YOLO V8 in mean average precision (mAP), achieving an exceptional mAP of 95.1%. This heightened precision is attributed to the integration of Neural Architecture Search (NAS) into the YOLO V8 NAS model, fine-tuning its architecture specifically for crowd counting tasks. It analyzes various networking models proposed in earlier studies for analyzing crowded scenes in public spaces. It emphasizes the potential of a hybrid model involving an IP camera module and a Deep Neural Network for effective crowd sensing. In this setup, the IP camera captures video footage, while the DNN detects individuals and assesses crowd density based on the count of people recognized. This approach presents an encouraging solution for real-time crowd monitoring and management in public environments. © 2024 IEEE.

关键词： Face recognition

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：