检索结果-内蒙古大学图书馆

Measuring the contribution of multiple model representations in detecting adversarial instances

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Steinberg, Daniel Munro, Paul Intelligent Systems Program University of Pittsburgh School of Computing and Information University of Pittsburgh

Deep learning models have been used for a wide variety of tasks. They are prevalent in computer vision, natural language processing, speech recognition, and other areas. While these models have worked well under many scenarios, it has been shown that they are vulnerable to adversarial attacks. This has led to a proliferation of research into ways that such attacks could be identified and/or defended against. Our goal is to explore the contribution that can be attributed to using multiple underlying models for the purpose of adversarial instance detection. Our paper describes two approaches that incorporate representations from multiple models for detecting adversarial examples. We devise controlled experiments for measuring the detection impact of incrementally utilizing additional models. For many of the scenarios we consider, the results show that performance increases with the number of underlying models used for extracting representations. Code is available at https://***/dstein64/multi-advdetect. Copyright © 2021, The Authors. All rights reserved.

关键词： Speech recognition

Scalable Numerical Embeddings for Multivariate Time Series: Enhancing Healthcare Data Representation Learning

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Huang, Chun-Kai Hsieh, Yi-Hsien Chien, Ta-Jung Chien, Li-Cheng Sun, Shao-Hua Su, Tung-Hung Kao, Jia-Horng Lin, Che Taiwan Data Science Degree Program NTU Taiwan Department of Electrical Engineering NTU Taiwan Division of Gastroenterology and Hepatology NTU Hospital Taiwan Hepatitis Research Center NTU Hospital Taiwan Graduate Institute of Clinical Medicine College of Medicine NTU Taiwan Center for Advanced Computing and Imaging in Biomedicine NTU Taiwan Smart Medicine and Health Informatics Program NTU Taiwan

Multivariate time series (MTS) data, when sampled irregularly and asynchronously, often present extensive missing values. Conventional methodologies for MTS analysis tend to rely on temporal embeddings based on timestamps that necessitate subsequent imputations, yet these imputed values frequently deviate substantially from their actual counterparts, thereby compromising prediction accuracy. Furthermore, these methods typically fail to provide robust initial embeddings for values infrequently observed or even absent within the training set, posing significant challenges to model generalizability. In response to these challenges, we propose SCAlable Numerical Embedding (SCANE), a novel framework that treats each feature value as an independent token, effectively bypassing the need for imputation. SCANE regularizes the traits of distinct feature embeddings and enhances representational learning through a scalable embedding mechanism. Coupling SCANE with the Transformer Encoder architecture, we develop the Scalable nUMerical eMbeddIng Transformer (SUMMIT), which is engineered to deliver precise predictive outputs for MTS characterized by prevalent missing entries. Our experimental validation, conducted across three disparate electronic health record (EHR) datasets marked by elevated missing value frequencies, confirms the superior performance of SUMMIT over contemporary state-of-the-art approaches addressing similar challenges. These results substantiate the efficacy of SCANE and SUMMIT, underscoring their potential applicability across a broad spectrum of MTS data analytical tasks. Copyright © 2024, The Authors. All rights reserved.

关键词： Embeddings

Visualizing representations of adversarially perturbed inputs

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Steinberg, Daniel Munro, Paul Intelligent Systems Program University of Pittsburgh School of Computing and Information University of Pittsburgh

It has been shown that deep learning models are vulnerable to adversarial attacks. We seek to further understand the consequence of such attacks on the intermediate activations of neural networks. We present an evaluation metric, POP-N, which scores the effectiveness of projecting data to N dimensions under the context of visualizing representations of adversarially perturbed inputs. We conduct experiments on CIFAR-10 to compare the POP-2 score of several dimensionality reduction algorithms across various adversarial attacks. Finally, we utilize the 2D data corresponding to high POP-2 scores to generate example visualizations. Copyright © 2021, The Authors. All rights reserved.

关键词： Deep learning

Understanding the Beneficial Role of Transition-Metal Layer Na⁺ Substitution on the Structure and Electrochemical Properties of the P2-Layered Cathode Na2+x Ni2-x/2TeO6

学校读者我要写书评

暂无评论

CHEMISTRY OF MATERIALS 2025年第9期37卷 3040-3053页

作者： Grundish, Nicholas S. Henkelman, Graeme Goodenough, John B. Delmas, Claude Carlier, Dany Seymour, Ieuan D. Univ Texas Austin Mat Sci & Engn Program Austin TX 78712 USA Univ Texas Texas Mat Inst Austin TX 78712 USA Univ Texas Austin Dept Chem Austin TX 78712 USA Univ Texas Austin Oden Inst Computat Engn & Sci Austin TX 78712 USA Univ Bordeaux Bordeaux INP CNRS ICMCBUMR 5026 F-33600 Pessac France CNRS FR 3459 Reseau Francaissur Stockage Electrochim Energie RS2E F-80039 Amiens France Univ Aberdeen Adv Ctr Energy & Sustainabil Sch Nat & Computing Sci Dept Chem Aberdeen AB24 3FX Scotland

Layered Na x MO2 sodium oxide positive electrode materials have experienced renewed interest owing to the current commercial attention on sodium-ion batteries. Although there are many attractive qualities of these materials, they suffer from serious shortcomings owing to Na+ ordering and transition-metal layer gliding that cause a plethora of voltage plateaus during cycling. The P2-layered Na2+x Ni2-x/2TeO6 (0 <= x <= 0.5) system provides a framework for investigating the effect of dual Na+ substitution into the sodium layer and the transition-metal layer of the structure and its effects on the electrochemical properties of the materials. A careful investigation into the synthesis and properties of these materials reveals that the sodium content used during material preparation has a drastic effect on the composition and electrochemical profile of these materials. The sodium substitution disrupts ordering within the transition-metal layer, thereby disrupting Na+ ordering in the adjacent sodium layers. Beyond a critical sodium concentration, the layer stacking shifts, and all voltage plateaus of the P2-Na2Ni2TeO6 material are no longer observed at 4.4 V versus Na+/Na. These results also question the common belief that additional sodium precursor is required when preparing layered sodium oxide cathodes, providing new guidelines for material synthesis and characterization.

关键词： Sodium-ion batteries

How Good is ChatGPT at Audiovisual Deepfake Detection: A Comparative Study of ChatGPT, AI Models and Human Perception

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Shahzad, Sahibzada Adil Hashmi, Ammarah Peng, Yan-Tsung Tsao, Yu Wang, Hsin-Min The Social Networks and Human-Centered Computing Program Taiwan International Graduate Program Academia Sinica Taipei Taiwan The Department of Computer Science National Chengchi University Taipei Taiwan The Institute of Information Systems and Applications National Tsing Hua University Hsinchu Taiwan The Research Center for Information Technology Innovation Academia Sinica Taipei Taiwan The Institute of Information Science Academia Sinica Taiwan

Multimodal deepfakes involving audiovisual manipulations are a growing threat because they are difficult to detect with the naked eye or using unimodal deep learning-based forgery detection methods. Audiovisual forensic models, while more capable than unimodal models, require large training datasets and are computationally expensive for training and inference. Furthermore, these models lack interpretability and often do not generalize well to unseen manipulations. In this study, we examine the detection capabilities of a large language model (LLM) (i.e., ChatGPT) to identify and account for any possible visual and auditory artifacts and manipulations in audiovisual deepfake content. Extensive experiments are conducted on videos from a benchmark multimodal deepfake dataset to evaluate the detection performance of ChatGPT and compare it with the detection capabilities of state-of-the-art multimodal forensic models and humans. Experimental results demonstrate the importance of domain knowledge and prompt engineering for video forgery detection tasks using LLMs. Unlike approaches based on end-to-end learning, ChatGPT can account for spatial and spatiotemporal artifacts and inconsistencies that may exist within or across modalities. Additionally, we discuss the limitations of ChatGPT for multimedia forensic tasks. © 2024, CC BY.

关键词： Audiovisual

Broadband Beamforming via Frequency Invariance Transformation and PARAFAC Decomposition for Jamming Mitigation in V2X Scenarios

学校读者我要写书评

暂无评论

Broadband Beamforming via Frequency Invariance Transformatio...

IEEE Conference on Vehicular Technology (VTC)

作者： Daniel V. de Lima João Paulo J. da Costa Antonio Arlis S. da Silva Giovanni A. Santos José Alfredo R. Vargas Auzuir R. de Alexandria Postgraduate Program in Telecommunications Engineering (PPGET) IFCE Fortaleza Brazil Hamm-Lippstadt University of Applied Sciences (HSHL) Lippstadt Germany Graduate School for Applied Research in North Rhine-Westphalia (PKNRW) Bochum Germany Karlsruhe Institute of Technology (KIT) Karlsruhe Gennany Graduate Program in Computing (PPGC) UFRGS Porto Alegre Brazil Department of Electrical Engineering UnB Brasilia Brazil

ISBN: (数字)9798350387414

ISBN: (纸本)9798350387421

Vehicle-to-Everything (V2X) communication can considerably improve the efficiency and safety of autonomous driving and advanced driver-assistance systems (ADASs). However, V2X communication can be considerably degraded in the presence of cyberattacks, such as radio jamming. Traditionally, beamforming techniques can be applied to increase the signal-to-interference plus noise ratio (SINR). This paper evaluates broadband beamforming in the mmWave spectrum against radio jamming in V2X communication. The exploitation of the mm Wave spectrum in 5G-V2X communication has a natural advantage against radio jamming. First, attenuation is stronger in the mmWave spectrum in the range of 40 GHz or higher than in the traditional 5.9 GHz. Second, to generate broadband radio jamming, the radio jammer requires much more complex hardware and energy consumption. Third, by using broadband beamforming, broadband radio jamming can be considerably attenuated, limiting the degradation of the spectrum by the radio jamming. According to our numerical results, gains of close to 30 dB SINR can be achieved. We propose a broad-band beamforming technique based on the canonical polyadic decomposition via generalized eigenvalue decomposition (CPD-GEVD). The CPD-GEVD broadband beamforming outperforms state-of-the-art beamforming algorithms in most V2X scenarios presented in this paper.

关键词： Vehicular and wireless technologies Array signal processing Interference Attenuation Broadband communication Jamming Millimeter wave communication

Comprehensive Evaluation of the Lakeside Café Point of Sale (POS) Application: A Synergistic Analysis Using SUS and EUCS Models

学校读者我要写书评

暂无评论

Comprehensive Evaluation of the Lakeside Café Point of Sale...

International Seminar on Intelligent Technology and Its Applications (ISITIA)

作者： Jonathan Arya Wibowo Rahmat Yasirandi Rio Guntur Utomo Rahmat Indra Pratama Anom School of Industrial Engineering Telkom University Bandung Indonesia Faculty of Information Technology King Mongkut’s Institute of Technology Ladkrabang Bangkok Thailand School of Computing Telkom University Bandung Indonesia Doctoral Program of Tourism Trisakti Institute of Tourism Jakarta Indonesia

ISBN: (数字)9798350378573

ISBN: (纸本)9798350378580

Lakeside Café, located in the Telkom University, processes approximately 80-100 drink transactions daily. Having used their current point-of-sale (POS) application for over a year, an urgent evaluation of its performance and user satisfaction is necessary to ensure ongoing efficiency. This study employs the System Usability Scale (SUS) model and the Extended Unified Theory of Acceptance and Use of Technology (EUCS) model. This combined approach enhances the understanding of the application's strengths and identifies areas for improvement, forming a solid foundation for evaluating user satisfaction. The study reveals notable usability of the Lakeside Café POS application, evidenced by an SUS score of 78.8, ranking it as "ACCEPTABLE" and "GOOD." The EUCS model analysis further highlights potential enhancements, with the Content and Functionality dimensions scoring 3.5 and 3.65, respectively, indicating room for workflow process development. These findings emphasize the need for improved content delivery and strengthened application functionality to boost user satisfaction and maximize system performance. This research enriches the dialogue on technological acceptance, providing valuable insights for developers and stakeholders focused on advancing POS applications. The use of SUS and EUCS models offers a comprehensive evaluation method, fostering continuous advancements and innovations in digital solutions within the food and beverage sector.

关键词： Seminars Analytical models Technological innovation System performance Solids Stakeholders Beverage industry

Multiomics Reveals Induction of Neuroblastoma SK-N-BE(2)C Cell Death by Mitochondrial Division Inhibitor 1 through Multiple Effects

学校读者我要写书评

暂无评论

Journal of Proteome Research 2024年第1期23卷 301-315页

作者： Wang, Wei-Hsuan Kao, Yi-Chun Hsieh, Chiao-Hui Tsai, Shin-Yu Cheung, Chantal Hoi Yin Huang, Hsuan-Cheng Juan, Hsueh-Fen Genome and Systems Biology Degree Program Academia Sinica and National Taiwan University Taipei 106 Taiwan Department of Life Science National Taiwan University Taipei 106 Taiwan Center for Computational and Systems Biology National Taiwan University Taipei 106 Taiwan Institute of Biomedical Informatics National Yang Ming Chiao Tung University Taipei 112 Taiwan Graduate Institute of Biomedical Electronics and Bioinformatics National Taiwan University Taipei 106 Taiwan Center for Advanced Computing and Imaging in Biomedicine National Taiwan University Taipei 106 Taiwan

Mitochondrial division inhibitor 1 (Mdivi-1) is a well-known synthetic compound aimed at inhibiting dynamin-related protein 1 (Drp1) to suppress mitochondrial fission, making it a valuable tool for studying mitochondrial dynamics. However, its specific effects beyond Drp1 inhibition remain to be confirmed. In this study, we employed integrative proteomics and phosphoproteomics to delve into the molecular responses induced by Mdivi-1 in SK-N-BE(2)C cells. A total of 3070 proteins and 1945 phosphorylation sites were identified, with 880 of them represented as phosphoproteins. Among these, 266 proteins and 97 phosphorylation sites were found to be sensitive to the Mdivi-1 treatment. Functional enrichment analysis unveiled their involvement in serine biosynthesis and extrinsic apoptotic signaling pathways. Through targeted metabolomics, we observed that Mdivi-1 enhanced intracellular serine biosynthesis while reducing the production of C24:1-ceramide. Within these regulated phosphoproteins, dynamic dephosphorylation of proteasome subunit alpha type 3 serine 250 (PSMA3-S250) occurred after Mdivi-1 treatment. Further site-directed mutagenesis experiments revealed that the dephosphorylation-deficient mutant PSMA3-S250A exhibited a decreased cell survival. This research confirms that Mdivi-1’s inhibition of mitochondrial division leads to various side effects, ultimately influencing cell survival, rather than solely targeting Drp1 inhibition. © 2023 American Chemical Society.

关键词： Mdivi-1 multiomics neuroblastoma phosphoproteome proteome targeted metabolome

Human Learners Integrate Visual and Linguistic Information in Cross-Situational Verb Learning 43

学校读者我要写书评

暂无评论

Human Learners Integrate Visual and Linguistic Information i...

43rd Annual Meeting of the Cognitive Science Society: Comparative Cognition: Animal Minds, CogSci 2021

作者： Zhang, Yayun Amatuni, Andrei Cain, Ellis Wang, Xizi Crandall, David Yu, Chen Department of Psychology University of Texas Austin United States Cognitive Science Program Indiana University Bloomington United States Luddy School of Informatics Computing and Engineering Indiana University Bloomington United States

Learning verbs is challenging because it is difficult to infer the precise meaning of a verb when there are a multitude of relations that one can derive from a single event. To study this verb learning challenge, we used children’s egocentric view collected from naturalistic toy-play interaction as learning materials and investigated how visual and linguistic information provided in individual naming moments as well as cross-situational information provided from multiple learning moments can help learners resolve this mapping problem using the Human Simulation Paradigm. Our results show that learners benefit from seeing children’s egocentric views compared to third-person observations. In addition, linguistic information can help learners identify the correct verb meaning by eliminating possible meanings that do not belong to the linguistic category. Learners are also able to integrate visual and linguistic information both within and across learning situations to reduce the ambiguity in the space of possible verb meanings. © Cognitive Science Society: Comparative Cognition: Animal Minds, CogSci *** rights reserved.

关键词： Toys