检索结果-内蒙古大学图书馆

arXiv 2021年

作者： Tariq, Shahroz Jeon, Sowon Woo, Simon S. Department of Computer Science and Engineering Sungkyunkwan University Suwon Korea Republic of Department of Applied Data Science Sungkyunkwan University Suwon Korea Republic of

Recently, significant advancements have been made in face recognition technologies using Deep Neural Networks. As a result, companies such as Microsoft, Amazon, and Naver offer highly accurate commercial face recognition web services for diverse applications to meet the end-user needs. Naturally, however, such technologies are threatened persistently, as virtually any individual can quickly implement impersonation attacks. In particular, these attacks can be a significant threat for authentication and identification services, which heavily rely on their underlying face recognition technologies’ accuracy and robustness. Despite its gravity, the issue regarding deepfake abuse using commercial web APIs and their robustness has not yet been thoroughly investigated. This work provides a measurement study on the robustness of black-box commercial face recognition APIs against Deepfake Impersonation (DI) attacks using celebrity recognition APIs as an example case study. We use five deepfake datasets, two of which are created by us and planned to be released. More specifically, we measure attack performance based on two scenarios (targeted and non-targeted) and further analyze the differing system behaviors using fidelity, confidence, and similarity metrics. Accordingly, we demonstrate how vulnerable face recognition technologies from popular companies are to DI attack, achieving maximum success rates of 78.0% and 99.9% for targeted (i.e., precise match) and non-targeted (i.e., match with any celebrity) attacks, respectively. Moreover, we propose practical defense strategies to mitigate DI attacks, reducing the attack success rates to as low as 0% and 0.02% for targeted and non-targeted attacks, respectively. Copyright © 2021, The Authors. All rights reserved.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

Enhancing Privacy and Security in Decentralized Social Systems: Blockchain-Based Approach

Enhancing Privacy and Security in Decentralized Social Syste...

引用

Cyber Resilience (ICCR), International Conference on

作者： Farah Abu-Dabaseh Mahmoud Alghizzawi Baker Ibrahim Alkhlaifat Abd Alrahman Ratib Ezmigna Amro Alzghoul Abdullah A.M AlSokkar Jassim Al-Gasawneh Computer Engineering Department Princess Sumaya University for Technology Amman Jordan Faculty of Business Applied Science Private University Amman Jordan Faculty of Business Al-Hussein Bin Talal University Ma'an Jordan Azman Hashim International Business School University Technology Malaysia Johor Malaysia Business Administration Department Amman Arab University Amman Jordan

ISBN: (数字)9798350394962

ISBN: (纸本)9798350394979

the emergence of decentralized social systems has provided multiple opportunities and challenges in the areas of privacy and security. Centralized platforms are sensitive to many vulnerabilities, including data breaches, illegal data sharing, and the possibility of censorship. Although decentralized platforms have been developed to address some concerns, there are challenges, especially in the areas of identity management and maintaining secure communication channels. Accordingly, this study aims to explore the possibility of using blockchain technology as a means of improving privacy and security in decentralized social networks. As the use of blockchain technology in the field of identity management, data encryption, and secure peer-to-peer communications, there are potential impacts and difficulties, especially in finding a good mix between openness and transparency. The results of our study show that the use of blockchain technology has a lot of potential to help solve many problems that already exist. However, it is important to be careful and use a well-thought-out plan while implementing it to make the most of its potential benefits. Finally, it discusses potential directions for future research and stresses how important it is to conduct more research in a field that is changing very rapidly.

关键词： Privacy Social networking (online) Communication channels Data breach Blockchains Peer-to-peer computing Encryption

来源：评论

学校读者我要写书评

暂无评论

Uncertainty estimation for time series classification Exploring predictive uncertainty in transformer-based models for variable stars

arXiv

引用

arXiv 2024年

作者： Cádiz-Leyton, Martina Cabrera-Vives, Guillermo Protopapas, Pavlos Moreno-Cartagena, Daniel Donoso-Oliva, Cristobal Becker, Ignacio Department of Computer Science Universidad de Concepción Edmundo Larenas 219 Concepción Chile John A. Paulson School of Engineering and Applied Science Harvard University CambridgeMA02138 United States Center for Data and Artificial Intelligence Universidad de Concepción Edmundo Larenas 310 Concepción Chile Nuncio Monseñor Sotero Sanz 100 Of. 104 Providencia Santiago Chile Chile Department of Computer Science Pontificia Universidad Catolica de Chile Macul Santiago7820436 Chile

Context. Classifying variable stars is key for understanding stellar evolution and galactic dynamics. With the demands of large astronomical surveys, machine learning models, especially attention-based neural networks, have become the state-of-the-art. While achieving high accuracy is crucial, enhancing model interpretability and uncertainty estimation is equally important to ensure that insights are both reliable and comprehensible. Aims. We aim to enhance transformer-based models for classifying astronomical light curves by incorporating uncertainty estimation techniques to detect misclassified instances. We tested our methods on labeled datasets from MACHO, OGLE-III, and ATLAS, introducing a framework that significantly improves the reliability of automated classification for the next-generation surveys. Methods. We used Astromer, a transformer-based encoder designed for capturing representations of single-band light curves. We enhanced its capabilities by applying three methods for quantifying uncertainty: Monte Carlo Dropout (MC Dropout), Hierarchical Stochastic Attention (HSA), and a novel hybrid method combining both approaches, which we have named Hierarchical Attention with Monte Carlo Dropout (HA-MC Dropout). We compared these methods against a baseline of deep ensembles (DEs). To estimate uncertainty estimation scores for the misclassification task, we selected Sampled Maximum Probability (SMP), Probability Variance (PV), and Bayesian Active Learning by Disagreement (BALD) as uncertainty estimates. Results. In predictive performance tests, HA-MC Dropout outperforms the baseline, achieving macro F1-scores of 79.8 ± 0.5 on OGLE, 84 ± 1.3 on ATLAS, and 76.6 ± 1.8 on MACHO. When comparing the PV score values, the quality of uncertainty estimation by HA-MC Dropout surpasses that of all other methods, with improvements of 2.5 ± 2.3 for MACHO, 3.3 ± 2.1 for ATLAS and 8.5 ± 1.6 for OGLE-III. Copyright © 2024, The Authors. All rights reserved.

关键词： Labeled data

来源：评论

学校读者我要写书评

暂无评论

Optimization and evaluation of motion sequences of an averaged human motion model by using kinematic techniques and new evaluation methods

Optimization and evaluation of motion sequences of an averag...

引用

作者： Kitzig, Andreas Demmer, Julia Naroska, Edwin Stockmanns, Gudrun Viga, Reinhard Grabmaier, Anton Niederrhein University of Applied Sciences Faculty of Electrical Engineering and Computer Science Ambient Intelligence Laboratory Reinarzstr. 49 Krefeld47805 Germany Niederrhein University of Applied Sciences Krefeld Germany University of Duisburg- Essen Department of Electronic Components and Circuits Duisburg Germany Duisburg Germany

Signal processing, pattern recognition as well as modelling and simulation require a large amount of reference data, both for the development of new methods and for their evaluation. Depending on the application, the availability of databases is rather low. In the field of biosignal processing with a focus on the functionalization of furniture for nursing and hospital facilities, a database from a motion capturing system (MoCap), and a method to generate averaged human motion sequences was presented in subsequent works by our research group. Evaluations revealed that the averaged motion sequences partly contain artifacts caused by the averaging and thus are not directly usable. To use the averaged motion sequences e.g., in simulation tasks, this paper presents an extension with kinematic methods to combine averaged motion sequences and to suppress and thus optimize inappropriate motion artifacts by error correction. To check whether the results are usable after the processing steps, four evaluation criteria are proposed. The evaluation of the resulting motion sequences shows that sequences are generated which do not fully correspond to human motion sequences but are well suited for simulation tasks. © 2021 by Walter de Gruyter Berlin/Boston.

关键词： Kinematics

来源：评论

学校读者我要写书评

暂无评论

Domain-Invariant Feature Learning for Domain Adaptation

引用

International Journal of Pattern Recognition and Artificial Intelligence 2023年第3期37卷

作者： Tu, Ching-Ting Lin, Hsiau-Wen Lin, Hwei Jen Tokuyama, Yoshimasa Chu, Chia-Hung Department of Applied Mathematics National Chung Hsing University Taichung401 Taiwan Department of Information Management Chihlee University of Technology Banqiao Dist. New Taipei City220305 Taiwan Department of Computer Science and Information Engineering Tamkang University Tamsui Dist. New Taipei City251 Taiwan Department of Media and Image Technology Faculty of Engineering Tokyo Polytechnic University Tokyo164-8678 Japan

Unsupervised domain adaptation (UDA) mainly explores how to learn domain-invariant features from the source domain when the target domain label is unknown. To learn domain-invariant features requires aligning the distribution of samples from two domains in the feature space, which can be achieved by minimizing the maximum mean discrepancy (MMD) of samples from the two domains. However, there is still no effective way to find the best parameter values of MMD. Such a problem is addressed in the MMD with deep kernels (MMD-D), whose optimal parameters can be obtained through training. This study proposes a method of domain-invariant feature learning for UDA, whose architecture, named MMDDCDA, comprises an MMD-D module and a cross domain adaptation (CDA) module. MMDDCDA performs alternating training similar to adversarial training to alternately boost the power of the two modules. To our knowledge, this is the first UDA method that performs such alternating training on a UDA architecture using MMD with deep kernels. Experimental validation showed that the proposed method yields state-of-the-art results among UDA methods using other MMD variants and some UDA benchmarks. © 2023 World Scientific Publishing Company.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Genetic Optimization Techniques for Enhancing Web Attacks Classification in Machine Learning

Genetic Optimization Techniques for Enhancing Web Attacks Cl...

引用

IEEE International Symposium on Dependable, Autonomic and Secure Computing (DASC)

作者： Ameera S. Jaradat Ahmad Nasayreh Qais Al-Na'amneh Hasan Gharaibeh Rabia Emhamed Al Mamlook Dep. of Computer Science Yarmouk University Irbid Jordan Dep. of Cyber Security and Cloud Computing Applied Science Private University Amman Jordan Dep. of Business Administration Trine University Indiana USA Department of Industrial Engineering University Zawia Tripoli Libia

Web-based applications are now the preferred approach for delivering a variety of services via the Internet. As a result of the globalization of commerce, web applications have been growing quickly and becoming increasingly complicated. Such applications have a significant security vulnerability in the online environment since they were developed with little experience and without testing or validation. Numerous attackers use this security vulnerability to take control of the program, modify the data, and steal the most crucial information. They may also access all internal, unauthorized items. This work presents a hybrid model that classifies website attacks as benign through the integration of four machine learning algorithms: gradient boost (GB), multilayer perceptron (MLP), logistic regression (LR) and K nearest neighbor (KNN). The work deployed Tree-based Pipeline Optimization Tool (TPOT) that utilizes the Genetic algorithm (GA) to extract the optimal parameter and consequently enhance the model performance. The model underwent evaluation utilizing a data set from the Canadian Institute 2023 that contains various types of attacks. Among these algorithms, GB achieved the best accuracy scores of 95%, 94% and 95% for accuracy, recall and F1-score, respectively.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Amplification and ellipticity enhancement of high-order harmonics in a neonlike x-ray laser dressed by an IR field

引用

Physical Review A 2023年第6期107卷 063511-063511页

作者： V. A. Antonov I. R. Khairulin M. Yu. Ryabikin M. A. Berrill V. N. Shlyaptsev J. J. Rocca Оlga Kocharovskaya Institute of Applied Physics of the Russian Academy of Sciences 46 Ulyanov Street Nizhny Novgorod 603950 Russia Lobachevsky State University of Nizhny Novgorod 23 Prospekt Gagarina Nizhny Novgorod 603950 Russia Department of Electrical and Computer Engineering Colorado State University Fort Collins Colorado 80523 USA Department of Physics Colorado State University Fort Collins Colorado 80523 USA Department of Physics and Astronomy and Institute for Quantum Science and Engineering Texas A&M University 578 University Drive College Station Texas 77843-4242 USA

In Khairulin et al. [Sci. Rep. 12, 6204 (2022)] we proposed a method for amplifying a train of sub-femtosecond pulses of circularly or elliptically polarized extreme ultraviolet radiation, composed by high-order harmonics of an infrared (IR) laser field, in a neonlike active medium of a plasma-based x-ray laser, additionally irradiated with a replica of a fundamental frequency laser field used to generate harmonics. Here we present the analytical theory and numerical study of an amplification process of an individual harmonic with either preservation of its polarization state or ellipticity enhancement. We also analyze in detail the spectral-temporal and polarization properties of a sub-femtosecond pulse train formed by a set of elliptically or circularly polarized harmonics during the propagation through the medium. We discuss also the possibility of an experimental implementation of the suggested technique in an active medium of an x-ray laser based on neonlike Ti12+ ions irradiated by an IR laser field with a wavelength of 3.2 µm.

关键词： Nonlinear optics Strong electromagnetic field effects Ultrafast optics X-ray beams & optics

来源：评论

学校读者我要写书评

暂无评论

Novel design of recurrent neural network for the dynamical of nonlinear piezoelectric cantilever mass–beam model

引用

European Physical Journal Plus 2024年第1期139卷 16-16页

作者： Kausar, Aneela Chang, Chuan-Yu Raja, Muhammad Asif Zahoor Zameer, Aneela Shoaib, Muhammad Graduate School of Engineering Science and Technology (Computer Science and Information Engineering) National Yunlin University of Science and Technology 123 University Road Section 3 Yunlin Douliou 64002 Taiwan Future Technology Research Center National Yunlin University of Science and Technology 123 University Road Section 3 Yunlin Douliou 64002 Taiwan Department of Computer and Information Sciences Pakistan Institute of Engineering and Applied Sciences Islamabad Nilore 45650 Pakistan AI Center Yuan Ze University Taoyüan 320 Taiwan

Piezoelectricity spans a wide range of fields, showcasing its practical significance. In the realm of consumer electronics, piezoelectric materials are utilized in buzzers, sensors, and actuators. These components find applications in everyday devices such as mobile phones, wearable gadgets, and home appliances. The aim of study is to present a novel design of layered recurrent neural network (LRNNs) for solving nonlinear piezoelectric cantilever mass–beam (PE-CMB) model representing the excitation frequency dynamics. The proposed PE-CMB model is mathematically represented with highly nonlinear second-order differential system. The synthetic dataset is created using Adams numerical scheme for piezoelectric cantilever beam system, and the data are randomly segmented for training samples to formulate the network and testing samples to record performance on unbiased data and validation samples for controllability of adaptive procedure of training the weights to approximate the solutions of PE-CMB with LRNNs. The efficiency and consistent performance of the LRNNs design are validated through an evaluation process encompassing measures such as achieved accuracy using mean squared error-derived cost functions, plots illustrating learning curves, adaptively controlled parameters, solution dynamics along with associated error analysis, and a statistical analysis involving distribution of error for each instances via histograms and regression metric. © 2024, The Author(s), under exclusive licence to Società Italiana di Fisica and Springer-Verlag GmbH Germany, part of Springer Nature.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Comparative Study of the Resolution of Ge-on-Si Photodetectors for $1\ \mu \mathrm{m}$ Infrared Signals

Comparative Study of the Resolution of Ge-on-Si Photodetecto...

引用

On-Line Conference (ONCON), Industrial Electronics Society Annual

作者： Mojtaba Jahangiri Paolo Sberna Amir Sammak Stoyan Nihtianov Department of Microelectronics Faculty of Electrical Engineering Mathematics and Computer Science Delft University of Technology Delft The Netherlands QuTech and Netherlands Organization for Applied Scientific Research (TNO) Delft The Netherlands

Most of the studies on narrow-band near-infrared detection reported so far are related to the $\mathbf{1}.\mathbf{3}\ \boldsymbol{\mu}\mathbf{m}$ and $\mathbf{1}.\mathbf{55}\ \boldsymbol{\mu}\mathbf{m}$ spectral windows. There is insufficient research work done on radiation detection in the narrow band around $1 \mu\mathrm{m}$ wavelength, which is just outside the Si $(\mathbf{0}.\mathbf{95}\ \boldsymbol{\mu} \mathbf{m})$ and GaAs $(\mathbf{0}.\mathbf{85}\ \boldsymbol{\mu} \mathbf{m})$ effective cut-off spectral sensitivity. This paper presents a p + n Ge-on-Si detector with a customized large active window, employing the PureGaB technology, to detect radiation in a very narrow band around $\mathbf{1}\ \boldsymbol{\mu} \mathbf{m}$ . The advantages of the proposed detector are: (1) CMOS-compatibility and micro-spectroscopic capability; (2) low dark current and high photoresponsivity, compared to similar devices reported in the literature; (3) enhanced sensitivity to weak radiation by realizing an ultra-shallow and very thin depletion region. These detectors can be good candidates for measuring the YAG laser radiation and measuring stray radiation in photolithography.

关键词： Industrial electronics Sensitivity Semiconductor lasers Current measurement Lithography Gallium arsenide Detectors

来源：评论

学校读者我要写书评

暂无评论

InGaN-based blue resonant cavity micro-LEDs with staggered multiple quantum wells enabling full-color and low-crosstalk micro-LED displays

Next Nanotechnology

引用

Next Nanotechnology 2024年 5卷

作者： Huang, Wei-Ta Lee, Tzu-Yi Bai, Yi-Hong Wang, Hsiang-Chen Hung, Yu-Ying Hong, Kuo-Bin Chen, Fang-Chung Lin, Chia-Feng Chang, Shu-Wei Han, Jung He, Jr-Hau Hong, Yu-Heng Kuo, Hao-Chung Department of Photonics Institute of Electro-Optical Engineering College of Electrical and Computer Engineering National Yang Ming Chiao Tung University Hsinchu 30010 Taiwan Semiconductor Research Center Hon Hai Research Institute Taipei 11492 Taiwan Department of Materials Science and Engineering National Chung Hsing University Taichung 402202 Taiwan Research Center for Applied Sciences Academia Sinica Taipei 11529 Taiwan Department of Electrical Engineering Yale University New Haven CT United States Department of Materials Science and Engineering City University of Hong Kong Tat Chee Avenue Kowloon Hong Kong

Herein, we proposed a unique structural design for indium gallium nitride (InGaN) based blue resonant cavity micro-light-emitting diodes (RC-μ-LEDs), focusing on the design, fabrication, and the relevant performance analyses. The proposed RC-μ-LEDs possess a three-layer staggered InGaN/GaN multiple quantum wells (MQWs) within the nanoporous Distributed Bragg Reflectors (NP-DBRs) and the conventional DBRs, introducing light confinement within such a resonant cavity. A passivation layer using atomic layer deposition (ALD) is adopted to reduce the leakage current from sidewall defects as well. Consequently, for the resulting RC-μ-LEDs, the divergence angle (DA) can be achieved down to 39.04°. While the input current increases from 1.77 A/cm² to 54 A/cm², the peak wavelength will shift from 456.16 nm to 449.18 nm, a blue shift of only 6.98 nm. Finally, we also discuss the temperature-dependent characteristics and the corresponding behaviors of our RC-μ-LEDs. Our demonstrated RC-μ-LEDs exhibit great wavelength stability with a diminished divergence angle, thus enabling full-color and low-crosstalk micro-LED displays for on-demand high-resolution applications. © 2024 The Authors

关键词： Color conversion InGaN Light-emitting diode Micro-LED Nanoporous DBR Quantum dot Resonant cavity Staggered quantum well Temperature dependent

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：