检索结果-内蒙古大学图书馆

Uncertainty quantification in large language models through convex hull analysis

Discover Artificial Intelligence 2024年第1期4卷 1-14页

作者： Catak, Ferhat Ozgur Kuzlu, Murat Department of Electrical Engineering and Computer Science University of Stavanger Rogaland Norway Department of Engineering Technology Old Dominion University NorfolkVA United States

Uncertainty quantification approaches have been more critical in large language models (LLMs), particularly high-risk applications requiring reliable outputs. However, traditional methods for uncertainty quantification, such as probabilistic models and ensemble techniques, face challenges when applied to the complex and high-dimensional nature of LLM-generated outputs. This study proposes a novel geometric approach to uncertainty quantification using convex hull analysis. The proposed method leverages the spatial properties of response embeddings to measure the dispersion and variability of model outputs. The prompts are categorized into three types, i.e., ’easy’, ’moderate’, and ’confusing’, to generate multiple responses using different LLMs at varying temperature settings. The responses are transformed into high-dimensional embeddings via a BERT model and subsequently projected into a two-dimensional space using Principal Component Analysis (PCA), Isomap, Multidimensional Scaling (MDS). The Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is utilized to cluster the embeddings and compute the convex hull for each selected cluster. The experimental results indicate that the uncertainty of the model for LLMs depends on the prompt complexity, the model, and the temperature setting. © The Author(s) 2024.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

Pruning techniques for artificial intelligence networks: a deeper look at their engineering design and bias: the first review of its kind

引用

Multimedia Tools and Applications 2025年第11期84卷 9591-9665页

作者： Mohanty, Lopamudra Kumar, Ashish Mehta, Vivek Agarwal, Mohit Suri, Jasjit S. School of Computer Science Engineering and Technology Bennett University Gr. Noida India Department of Computer Science ABES Engineering College UP Ghaziabad India Stroke Diagnostic and Monitoring Division AtheroPoint™ RosevilleCA United States

Trained Artificial Intelligence (AI) models are challenging to install on edge devices as they are low in memory and computational power. Pruned AI (PAI) models are therefore needed with minimal degradation in performance. This is the first study of its kind where we have characterized three types of PAI paradigms, namely, weight pruning (WP), channel pruning (CP), and hybrid pruning (HP). Further, the study also analyses the risk of bias (RoB) in PAI models, and explainable PAI (XPAI) networks. PRISMA model was adapted for the selection of the best 145 studies. The review had two hypotheses: (i) PAI had bias and (ii) the order of performance for the three PAI is: HP > CP > WP in terms of performance. Thirty-two AI attributes using each of two experts were categorized into four clusters, namely, architecture, optimization, dataset design, and scientific validation. The RoB was analyzed in these PAI studies using AP(ai)Bias 2.0 (AtheroPoint™, CA, USA), which was based on a scoring-and-ranking strategy. This was then compared against the Butterfly model, composed of radial and regional biases, and analyzed using a Venn diagram in three independent bins based on low-moderate and moderate-high cut-offs. Recommendations were laid out for low-bias studies. PAI studies that lack adequate AI attributes were more inclined to the RoB. It has been observed that 19.64%, 21.42%, and 58.92% studies were in low-bias, moderate-bias, and high-bias bins, respectively. Pruning made inference faster and reduced sparsity. We have observed low participation of the PAI studies in the XPAI framework. The study presented the challenges and five-step recommendations for mitigating bias in PAI models. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Risk assessment

来源：评论

学校读者我要写书评

暂无评论

A systematic literature review of visual feature learning: deep learning techniques, applications, challenges and future directions

引用

Multimedia Tools and Applications 2024年 1-58页

作者： Abdullahi, Mohammed Oyelade, Olaide Nathaniel Kana, Armand Florentin Donfack Bagiwa, Mustapha Aminu Abdullahi, Fatimah Binta Junaidu, Sahalu Balarabe Iliyasu, Ibrahim Ore-ofe, Ajayi Chiroma, Haruna Department of Computer Science Ahmadu Bello University Zaria Zaria Nigeria Department of Mechanical Engineering Ahmadu Bello University Zaria Nigeria Department of Computer Engineering Ahmadu Bello University Zaria Nigeria College of Computer Science and Engineering University of Hafr Al Batin Hafar Al-Batin31991 Saudi Arabia School of Electronics Electrical Engineering and Computer Science Queen’s University Belfast Belfast United Kingdom

Visual Feature Learning (VFL) is a critical area of research in computer vision that involves the automatic extraction of features and patterns from images and videos. The applications of VFL are vast, including object detection and recognition, facial recognition, scene understanding, medical image analysis, and autonomous vehicles. In this paper, we propose to conduct extensive systematic literature review (SLR) on VFL based on deep learning algorithms. The paper conducted an SLR covering deep learning algorithms such as Convolutional Neural Networks (CNNs), Autoencoders, and Generative Adversarial Networks (GANs) including their variants. The review highlights the importance of VFL in computer vision and the limitations of traditional feature extraction techniques. Furthermore, it provides an in-depth analysis of the strengths and weaknesses of various deep learning algorithms for solving problems in VFL. The discussion of the applications of VFL provides an insight into the impact of VFL on various industries and domains. The review also analyzed the challenges faced by VFL, such as data scarcity and quality, overfitting, generalization, interpretability, and explainability. The discussion of future directions for VFL includes hybrid techniques, unsupervised feature learning, continual learning, attention-based models, and explainable AI. These techniques aim to address the challenges faced by VFL and improve the performance of the models. The systematic literature review concludes that VFL is a rapidly evolving field with the potential to transform many industries and domains. The review highlights the need for further research in VFL and emphasizes the importance of responsible use of VFL models in various applications. The review provides valuable insights for researchers and practitioners in the field of computer vision, who can use these insights to enhance their work and ensure the responsible use of VFL models. © The Author(s), under exclusive licence to

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

Jamming Power Optimization for Data Freshness in Covert Relay Networks 14

Jamming Power Optimization for Data Freshness in Covert Rela...

引用

14th International Conference on Information and Communication Technology Convergence, ICTC 2023

作者： Sim, Seungmin Kim, Jinwoong Lee, Jemin Department of Electrical and Computer Engineering Korea Republic of Department of Electrical Engineering and Computer Science Korea Republic of

ISBN: (纸本)9798350313277

In this paper, we analyze covert amplify-and-forward (AF) relay networks with a metric for measuring the data freshness, i.e, age of information (AoI), with aid of the cooperative jammer that generates artificial noise (AN) to protect from Warden Willie's surveillance. We first derive lower boundary of covertness, average AoI and formulate an average AoI minimization problem by optimizing the jamming power of the cooperative jammer. Our analysis shows that the optimal jamming power is the minimum value that satisfies the lower boundary of covertness, which varies depending on the prior probability. © 2023 IEEE.

关键词： AF age of information Covert communication detection error proability relaying network

来源：评论

学校读者我要写书评

暂无评论

Regression analysis of squeezing-induced hybrid nanofluid flow in Darcy-Forchheimer porous medium

引用

Applied Mathematics and Mechanics(English Edition) 2025年第1期46卷 193-208页

作者： K.MUHAMMAD M.SARFRAZ Department of Humanities and Sciences School of Electrical Engineering and Computer ScienceNational University of Sciences and TechnologyIslamabad 44000Pakistan Department of Mathematics Quaid-i-Azam UniversityIslamabad 44000Pakistan

This article presents a mathematical model addressing a scenario involving a hybrid nanofluid flow between two infinite parallel *** plate remains stationary,while the other moves downward at a squeezing *** space between these plates contains a Darcy-Forchheimer porous medium.A mixture of water-based fluid with gold(Au)and silicon dioxide(Si O2)nanoparticles is *** contrast to the conventional Fourier's heat flux equation,this study employs the Cattaneo-Christov heat flux equation.A uniform magnetic field is applied perpendicular to the flow direction,invoking magnetohydrodynamic(MHD)***,the model accounts for Joule heating,which is the heat generated when an electric current passes through the *** problem is solved via NDSolve in *** and statistical analyses are conducted to provide insights into the behavior of the nanomaterials between the parallel plates with respect to the flow,energy transport,and skin *** findings of this study have potential applications in enhancing cooling systems and optimizing thermal management *** is observed that the squeezing motion generates additional pressure gradients within the fluid,which enhances the flow rate but reduces the frictional ***,the fluid is pushed more vigorously between the plates,increasing the flow *** the fluid experiences higher flow rates due to the increased squeezing effect,it spends less time in the region between the *** thermal relaxation,however,abruptly changes the temperature,leading to a decrease in the temperature fluctuations.

关键词： convective boundary condition Darcy-Forchheimer medium hybrid nanofuid Joule heating magnetohydrodynamic(MHD) numerical solution squeezing flow regression analysis

来源：评论

学校读者我要写书评

暂无评论

Enhanced Cybercrime Detection on Twitter Using Aho-Corasick Algorithm and Machine Learning Techniques

引用

Informatica (Slovenia) 2024年第18期48卷 97-108页

作者： Rawat, Romil Raj, A. Samson Arun Chakrawarti, Rajesh Kumar Sankaran, Krishnan Sakthidasan Sarangi, Sanjaya Kumar Rawat, Hitesh Rawat, Anjali Department of Computer Science Engineering Shri Vaishnav Institute of Information Technology Shri Vaishnav Vidyapeeth Vishwavidyalaya Indore India Division of Computer Science and Engineering School of Computer Science and Technology Karunya Institute of Technology and Sciences Tamil Nadu Coimbatore India MP Indore India Department od Electronics and Communication Engeneering Hindustan Institute of Technology and Science Chennai India Department of Computer Science and Engineering Utkal University Bhubaneswar India Department of Business Management and Economics University of Extremadura Spain Department of Computer and Communication Technology University of Extremadura Spain

The proposed work objective is to adapt Online social networking (OSN) is a type of interactive computer-mediated technology that allows people to share information through virtual networks. The microblogging feature of Twitter makes cyberspace prominent (usually accessed via the dark web). The work used the datasets and considered the Scrape Twitter Data (Tweets) in Python using the SN-Scrape module and Twitter 4j API in JAVA to extract social data based on hashtags, which is used to select and access tweets for dataset design from a profile on the Twitter platform based on locations, keywords, and hashtags. The experiments contain two datasets. The first dataset has over 1700 tweets with a focus on location as a keypoint (hacking-for-fun data, cyber-violence data, and vulnerability injector data), whereas the second dataset only comprises 370 tweets with a focus on reposting of tweet status as a keypoint. The method used is focused on a new system model for analysing Twitter data and detecting terrorist attacks. The weights of susceptible keywords are found using a ternary search by the Aho-Corasick algorithm (ACA) for conducting signature and pattern matching. The result represents the ACA used to perform signature matching for assigning weights to extracted words of tweet. ML is used to evaluate Twitter data for classifying patterns and determining the behaviour to identify if a person is a terrorist. SVM (Support Vector Machine) proved to be a more accurate classifier for predicting terrorist attacks compared to other classifiers (KNN- K-Nearest Neighbour and NB-Naïve Bayes). The 1st dataset shows the KNN-Acc. -98.38% and SVM Accuracy as 98.85%, whereas the 2nd dataset shows the KNN-Acc. -91.68% and SVM Accuracy as 93.97%. The proposed work concludes that the generated weights are classified (cyber-violence, vulnerability injector, and hacking-for-fun) for further feature classification. Machine learning (ML) [KNN and SVM] is used to predict the occurrence and

关键词： Tweets

来源：评论

学校读者我要写书评

暂无评论

Inductive Lottery Ticket Learning for Graph Neural Networks

引用

Journal of computer science & Technology 2024年第6期39卷 1223-1237页

作者： Yong-Duo Sui Xiang Wang Tianlong Chen Meng Wang Xiang-Nan He Tat-Seng Chua School of Data Science University of Science and Technology of ChinaHefei 230027China Department of Electrical and Computer Engineering The University of Texas at AustinAustin 78712U.S.A School of Computer Science and Information Engineering Hefei University of TechnologyHefei 230009China School of Computing National University of SingaporeSingapore

Graph neural networks (GNNs) have gained increasing popularity, while usually suffering from unaffordable computations for real-world large-scale applications. Hence, pruning GNNs is of great need but largely unexplored. The recent work Unified GNN Sparsification (UGS) studies lottery ticket learning for GNNs, aiming to find a subset of model parameters and graph structures that can best maintain the GNN performance. However, it is tailed for the transductive setting, failing to generalize to unseen graphs, which are common in inductive tasks like graph classification. In this work, we propose a simple and effective learning paradigm, Inductive Co-Pruning of GNNs (ICPG), to endow graph lottery tickets with inductive pruning capacity. To prune the input graphs, we design a predictive model to generate importance scores for each edge based on the input. To prune the model parameters, it views the weight’s magnitude as their importance scores. Then we design an iterative co-pruning strategy to trim the graph edges and GNN weights based on their importance scores. Although it might be strikingly simple, ICPG surpasses the existing pruning method and can be universally applicable in both inductive and transductive learning settings. On 10 graph-classification and two node-classification benchmarks, ICPG achieves the same performance level with 14.26%–43.12% sparsity for graphs and 48.80%–91.41% sparsity for the GNN model.

关键词： lottery ticket hypothesis graph neural networks neural network pruning

来源：评论

学校读者我要写书评

暂无评论

Optimized Consensus for Blockchain in Internet of Things Networks via Reinforcement Learning

引用

Tsinghua science and Technology 2023年第6期28卷 1009-1022页

作者： Yifei Zou Zongjing Jin Yanwei Zheng Dongxiao Yu Tian Lan Institute of Intelligent Computing School of Computer Science and TechnologyShandong UniversityQingdao 266237China Department of Electrical and Computer Engineering George Washington UniversityWashingtonDC 20052USA

Most blockchain systems currently adopt resource-consuming protocols to achieve consensus between miners;for example,the Proof-of-Work(PoW)and Practical Byzantine Fault Tolerant(PBFT)schemes,which have a high consumption of computing/communication resources and usually require reliable communications with bounded ***,these protocols may be unsuitable for Internet of Things(IoT)networks because the IoT devices are usually lightweight,battery-operated,and deployed in an unreliable wireless ***,this paper studies an efficient consensus protocol for blockchain in IoT networks via reinforcement ***,the consensus protocol in this work is designed on the basis of the Proof-of-Communication(PoC)scheme directly in a single-hop wireless network with unreliable communications.A distributed MultiAgent Reinforcement Learning(MARL)algorithm is proposed to improve the efficiency and fairness of consensus for miners in the blockchain *** this algorithm,each agent uses a matrix to depict the efficiency and fairness of the recent consensus and tunes its actions and rewards carefully in an actor-critic framework to seek effective *** results from the simulation show that the fairness of consensus in the proposed algorithm is guaranteed,and the efficiency nearly reaches a centralized optimal solution.

关键词： consensus in blockchain Proof-of-Communication(PoC) MultiAgent Reinforcement Learning(MARL) Internet of Things(IoT)networks

来源：评论

学校读者我要写书评

暂无评论

A Unified Multi-Domain Face Normalization Framework for Cross-domain Prototype Learning and Heterogeneous Face Recognition

引用

IEEE Transactions on Information Forensics and Security 2025年 20卷 5282-5295页

作者： Pang, Meng Zhang, Wenjun Lu, Yang Cheung, Yiu-Ming Zhou, Nanrun Nanchang University School of Mathematics and Computer Sciences China Xiamen University Department of Computer Science and Technology School of Informatics China Hong Kong Baptist University Department of Computer Science Hong Kong Hong Kong Shanghai University of Engineering Science School of Electronic and Electrical Engineering China

Face normalization is a critical technique for improving the robustness and generalizability of face recognition systems by reducing intra-personal variations arising from expressions, poses, occlusions, illuminations, and domain shifts. Existing normalization methods, however, often lack the flexibility to handle multi-factorial variations and exhibit limited cross-domain adaptability. To address these challenges, we propose a Unified Multi-Domain Face Normalization Network (UMFN), which is designed to process facial images with diverse variations from various domains and reconstruct frontal, neutralized facial prototypes in the target domain. As an unsupervised domain adaptation model, the UMFN facilitates concurrent training across multiple cross-domain datasets and demonstrates robust prototype reconstruction capabilities. Notably, the UMFN functions as a joint prototype and feature learning framework, extracting domain-agnostic identity features through a decoupling mapping network and adversarial training with a feature domain classifier. Furthermore, we design an efficient Heterogeneous Face Recognition (HFR) network that integrates these domain-agnostic features and the identity-discriminative features extracted from normalized prototypes, enhanced by contrastive learning to improve identity recognition accuracy. Empirical evaluation on multiple cross-domain benchmark datasets validate the effectiveness of the UMFN for face normalization and the superiority of the HFR network for heterogeneous face recognition. © 2005-2012 IEEE.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Robust Wind-Resistant Hovering Control of Quadrotor UAVs Using Deep Reinforcement Learning

IEEE Transactions on Intelligent Vehicles

引用

IEEE Transactions on Intelligent Vehicles 2023年 1-10页

作者： Xue, Jun Liu, Ziniu Liu, Guanjun Zhou, Ziyuan Zhang, Kaiwen Tang, Ying Wang, Jiacun Department of Control Science and Engineering Tongji University Shanghai China Department of Computer Science Tongji University Shanghai China Department of Electrical and Computer Engineering Rowan University Glassboro NJ USA Department of Computer Science and Software Engineering Monmouth University West Long Branch NJ USA

Unmanned Aerial Vehicles (UAVs) have extensive applications such as logistics transportation and aerial photography. However, UAVs are sensitive to winds. Traditional control methods, such as proportional- integral-derivative controllers, generally fail to work well when the strength and direction of winds are changing frequently. In this work deep reinforcement learning algorithms are combined with a domain randomization method to learn robust wind-resistant hovering policies. A novel reward function is designed to guide learning. This reward function uses a constant reward to maintain a continuous flight of a UAV as well as a weight of the horizontal distance error to ensure the stability of the UAV at altitude. A five-dimensional representation of actions instead of the traditional four dimensions is designed to strengthen the coordination of wings of a UAV. We theoretically explain the rationality of our reward function based on the theories of Q-learning and reward shaping. Experiments in the simulation and real-world application both illustrate the effectiveness of our method. To the best of our knowledge, it is the first paper to use reinforcement learning and domain randomization to explore the problem of robust wind-resistant hovering control of quadrotor UAVs, providing a new way for the study of wind-resistant hovering and flying of UAVs. IEEE

关键词： Heuristic algorithms

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：