检索结果-内蒙古大学图书馆

Solving two-stage stochastic integer programs via representation learning

NEURAL NETWORKS 2025年 188卷 107446页

作者： Wu, Yaoxin Cao, Zhiguang Song, Wen Zhang, Yingqian Eindhoven Univ Technol Dept Ind Engn & Innovat Sci NL-5600 MB Eindhoven Netherlands Singapore Management Univ Singapore Sch Comp & Informat Syst Singapore 178902 Singapore Shandong Univ Inst Marine Sci & Technol Qingdao 266237 Peoples R China

Solving stochastic integer programs (SIPs) is extremely intractable due to the high computational complexity. To solve two-stage SIPs efficiently, we propose a conditional variational autoencoder (CVAE) for scenario representation learning. A graph convolutional network (GCN) based VAE embeds scenarios into a low-dimensional latent space, conditioned on the deterministic context of each instance. With the latent representations of stochastic scenarios, we perform two auxiliary tasks: objective prediction and scenario contrast, which predict scenario objective values and the similarities between them, respectively. These tasks further integrate objective information into the representations through gradient backpropagation. Experiments show that the learned scenario representations can help reduce scenarios in SIPs, facilitating high-quality solutions in a short computational time. This superiority generalizes well to instances of larger sizes, more scenarios, and various distributions.

关键词： Stochastic integer programs conditional variational autoencoder Semi-supervised learning Contrastive learning

来源：评论

学校读者我要写书评

暂无评论

Improving Local Fidelity and Interpretability of LIME by Replacing Only the Sampling Process With CVAE

引用

IEEE ACCESS 2025年 13卷 53084-53099页

作者： Yasui, Daisuke Sato, Hiroshi Natl Def Acad Dept Comp Sci Yokosuka Kanagawa 2390811 Japan

Knowing the basis of decisions is essential for using Machine Learning (ML) in various regions, such as medical diagnosis, finance, and organizational decision-making. The Local Interpretable Model agnostic Explanation (LIME) algorithm is an Explainable AI (XAI) method that can be applied to many black -box models. However, there is a problem in that the local fidelity of the interpretable model decreases. Therefore, local fidelity must be increased, and the more interpretability of the local surrogate model, the better. Moreover, the fewer the changes, the easier it is to apply them to other studies and the more significant the contribution to the XAI area. Therefore, there are studies that improve fidelity by changing only the sampling. However, no studies have been identified that improve local fidelity and interpretability by modifying only the sampling. To this end, we use a conditional variational autoencoder (CVAE) to modify the sampling of LIME in the classification task to improve local fidelity and interpretability. We conducted a study to improve local fidelity and interpretability by modifying only the sampling part and clarifying the mechanism behind this improvement. Results show that the proposed method improves local fidelity while increasing interpretability compared to existing methods and the factors contributing to this improvement. We also conducted experiments on stability, stability against noise, and execution time, and confirmed that the proposed method performs well, although not as well as methods specific to these issues. Our code is publicly released on GitHub (https://***/YasuiDaisuke/***) under the Apache 2.0 License.

关键词： Closed box autoencoders Training data Predictive models Linear regression Hands Adaptation models Training Stability criteria Neural networks LIME local fidelity interpretability conditional variational autoencoder

来源：评论

学校读者我要写书评

暂无评论

Enhancing choice-set generation and route choice modeling with data- and knowledge-driven approach

引用

TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES 2024年 162卷

作者： Liu, Dongjie Li, Dawei Gao, Kun Song, Yuchen Zhang, Tong Southeast Univ Sch Transportat Sipailou 2 Nanjing 210096 Peoples R China Chalmers Univ Technol Dept Architecture & Civil Engn SE-41296 Gothenburg Sweden Southeast Univ Jiangsu Key Lab Urban ITS Sipailou 2Xuanwu Dist Nanjing 210096 Peoples R China Jiangsu Prov Collaborat Innovat Ctr Modern Urban T Sipailou 2 Nanjing 210096 Peoples R China

Two central and interconnected problems arise in the specification of a "complete" path-based route choice model: choice-set generation and choice from a choice set. Choice-set generation poses a significant challenge in personalization and the enumeration of the full choice set with large size. Despite the continued prevalence of classic econometric models for modeling choices within a given set, this requirement of knowledge-driven modeling necessitates explicit model structures and intricate domain knowledge, which may result in practical biases. In this study, a conditional variational autoencoder (CVAE)-based choice set generation model is developed, which approximates the probability distribution of the underlying choice set generation process conditional on individual and OD characteristics without relying on expert knowledge. In order to facilitate a friendly integration between knowledge-driven econometric and machine learning approaches, a neural-embedded route choice model (IAP-NERCM) with implicit availability/ perception (IAP) of choice alternatives is proposed to automatically capture the heterogeneity of taste parameters without assuming any a priori relationship. Results based on synthetic data show that the proposed models are capable of reproducing the pre-defined coefficients. Field data of GPS data collected in Toyota City is used to future test the proposed models compared to classical statistical models. Results indicate that IAP-NERCM exhibits the ability to recover underlying taste function and achieves the best performance in terms of goodness-of-fit, predictability, and estimation time.

关键词： Route choice modeling Choice-set generation conditional variational autoencoder Implicit availability/perception Data and model-driven choice model

来源：评论

学校读者我要写书评

暂无评论

Parametrische Modellierung und generatives tiefes Lernen für den Brückenentwurf

引用

BAUTECHNIK 2024年第3期101卷 174-180页

作者： Kraus, Michael A. Kuhn, Sophia V. Hodel, Anna Bischof, Rafael Maissen, Alessandro Salamanca Mino, Luis Perez-Cruz, Fernando Swiss Fed Inst Technol Inst Baustat & Konstrukt Prof Massiv & Bruckenbau Stefano Franscini Platz 5 CH-8093 Zurich Switzerland Andreasturm 18th FloorAndreasstr 5 CH-8050 Zurich Switzerland

In Anbetracht der erheblichen Umweltauswirkungen des Bauwesens wird die Analyse und v. a. Optimierung der Nachhaltigkeit von Strukturen unter Beibehaltung des etablierten Zuverlassigkeitsniveaus immer wichtiger. Im Hochbausektor existieren erste Werkzeuge zur Lebenszyklusanalyse, diese sind jedoch nicht direkt ubertragbar auf Bruckentragwerke. Dieser Beitrag fasst die wesentlichen Ansatze und Ergebnisse von aktuellen Forschungsprojekten der Autoren an der ETH zusammen und erlautert insbesondere einen neuen Deep-Learning-basierten Ansatz zur Erkundung und Modellierung des Entwurfsraums parametrischer Bruckenmodelle und deren Leistungsbewertungen und veranschaulicht die Anwendung fur eine Mehrzieloptimierung von Stahlbetonrahmenbrucken. Zunachst werden Daten unter Verwendung eines parametrischen Bruckenmodells sowie der Ankoppelung von Analysesoftware synthetisch generiert und anschliessend bedingte variationelle autoencoder (CVAE) als Metamodell trainiert. Der CVAE dient im Rahmen des konzeptionellen Bruckenentwurfs als effizienter Co-Pilot sowohl fur die Vorwarts- als auch Ruckwartsanalyse. Die mit dem CVAE durchgefuhrte Sensitivitatsanalyse zeigt Beziehungen zwischen Entwurfsparametern und/oder Leistungskenngrossen sowie Optimierungspotenziale auf. Das hier vorgestellte integrierte Framework besitzt das Potenzial zur Realisierung einer effizienten Bruckenplanung unter insbesondere den Kriterien der Nachhaltigkeit und Tragsicherheit und kann problemlos auf andere parametrische Fragestellungen erweitert werden. Parametric modeling and generative deep learning for bridge designGiven the significant environmental impact of the construction industry, the analysis and, above all, optimization of the sustainability of structures while maintaining established levels of reliability are becoming increasingly important. While there are initial tools for life cycle analysis in the building sector, these are not directly transferable to bridge structures. This paper introduces

关键词： Bruckenbau kunstliche Intelligenz generativer Entwurf bedingte variationelle autoencoder bridge design artificial intelligence generative design conditional variational autoencoder

来源：评论

学校读者我要写书评

暂无评论

An efficient faults and attacks categorization model in IoT-based cyber physical systems using Dilated CNN and BiLSTM with multi-scale dense Attention module

引用

BIOMEDICAL SIGNAL PROCESSING AND CONTROL 2024年第PartB期96卷

作者： Laythkhaleel, Riyam Ibrahim, Abdullahi Abdu Naseri, Raghda Awad Shaban Farhan, Hameed Mutlag Altinbas Univ Elecr & Comp Engn Mahmutbey Dilmenler Caddesi 26 TR-34217 Istanbul Turkiye Tikrit Univ Dent Dept MMH4 876 Tikrit Iraq

The physical process with the digital computing channel and electronic computing is integrated by the Cyber-Physical Systems (CPS). The abnormality and failures are the two main sources that affect the performance of the CPS. In the CPS, the research on security analysis and fault diagnosis has attracted lots of interest from the researcher. However, the existing model does not find the difference between the fault and attacks, the existing approaches need adequate development for identifying the difference between the fault and attack in the Internet of Things (IoT)-CPS. In this research, a deep learning-based approach is developed to detect the attacks and faults in the IoT-CPS for enhancing the security of the network. The IoT-based data is garnered from a standard online source. After collecting the data, the extraction of the deep features is performed using the conditional variational autoencoder (CVA). The attained deep attributes are further taken for the weighted feature fusion process in which the required weights are chosen in an optimal manner using the Enhanced Egret Swarm Optimization (EESO) algorithm. The obtained weighted fused features are inputted into the Dilated Convolutional Neural Network and Bidirectional Long Short-Term Memory with Multi-Scale Dense Attention (DCNN-Bi-LSTM-MSDA). The classification outcomes are obtained from the DCNN-Bi-LSTM-MSDA module. Throughout the result analysis, the accuracy and NPV rate of the designed model is 94.16% and 99.38%. The validation of the fault and attack classification offered by the implemented deep learning-oriented failure and abnormality classification scheme in IoT-based CPS is done against several traditional models.

关键词： Internet of Things Cyber-Physical System Fault and Attack Classification conditional variational autoencoder Enhanced Egret Swarm Optimization Algorithm Dilated Convolutional Neural Network Bidirectional Long Short-Term Memory with Multi-Scale Dense Attention

来源：评论

学校读者我要写书评

暂无评论

Synthetic and privacy-preserving traffic trace generation using generative AI models for training Network Intrusion Detection Systems

引用

JOURNAL OF NETWORK AND COMPUTER APPLICATIONS 2024年 229卷

作者： Aceto, Giuseppe Giampaolo, Fabio Guida, Ciro Izzo, Stefano Pescape, Antonio Piccialli, Francesco Prezioso, Edoardo Univ Naples Federico II Dept Elect Engn & Informat Technol DIETI Via Claudio 21 I-80125 Naples Italy Univ Naples Federico II Dept Math & Applicat R Caccioppoli DMA Via Cinthia I-80126 Naples Italy Univ Bergamo Via Caniana 2 I-24127 Bergamo Italy

Network Intrusion Detection Systems (NIDS) are crucial tools for protecting networked devices from cyberattacks. Recent development in the field of Artificial Intelligence (AI) has provided tremendous advantages in implementing NIDSs able to monitor network traffic and block cyberattacks in real-time. In the literature, it is widely recognized that the effective training of a NIDS requires a large quantity of labeled traffic, representative of attacks. Nonetheless, the availability of public and abundant datasets remains remarkably restricted due to the cost of gathering and labeling real traffic traces and privacy concerns for sharing them. To tackle these challenges, in this paper we present a generative AI model capable of synthesizing anonymized traffic traces from real ones, thus dealing with privacy, abundance, and representativeness. The proposal is based on a conditional variational autoencoder (CVAE) and a preprocessing procedure specifically designed for the generation of new traffic traces. To validate our solution, we conduct an extensive empirical study leveraging three recent and publicly -available datasets, containing benign and malicious traffic. The validation is carried out from both the perspectives of classification performance of a robust NIDS and the quality of synthetic data, in comparison to the utilization of real data. We compare our CVAE with two state-of-the-art AI -based traffic data generators and prove that, trained with traces emitted by our generative model, a NIDS has a limited F1score loss compared to training on real data;competing models instead struggle or fail to generate traces that are as effective for NIDS training and as statistically similar to the original. We make the synthetic datasets available in both PCAP and tabular formats, to facilitate the reproducibility of our findings and encourage further exploration in the field of generative AI for networking.

关键词： Network Intrusion Detection System User privacy Traffic trace generation Traffic malware classification Generative artificial intelligence conditional variational autoencoder

来源：评论

学校读者我要写书评

暂无评论

EmoMusicTV: Emotion-Conditioned Symbolic Music Generation With Hierarchical Transformer VAE

引用

IEEE TRANSACTIONS ON MULTIMEDIA 2024年 26卷 1076-1088页

作者： Ji, Shulei Yang, Xinyu Xi An Jiao Tong Univ Sch Comp Sci & Technol Xian 710049 Peoples R China

Emotion is one of the most crucial attributes of music. However, due to the scarcity of emotional music datasets, emotion-conditioned symbolic music generation using deep learning techniques has not been investigated in depth. In particular, no study explores conditional music generation with the guidance of emotion, and few studies adopt time-varying emotional conditions. To address these issues, first, we endow three public lead sheet datasets with fine-grained emotions by automatically computing the valence labels from the chord progressions. Second, we propose a novel and effective encoder-decoder architecture named EmoMusicTV to explore the impact of emotional conditions on multiple music generation tasks and to capture the rich variability of musical sequences. EmoMusicTV is a transformer-based variational autoencoder (VAE) that contains a hierarchical latent variable structure to model holistic properties of the music segments and short-term variations within bars. The piece-level and bar-level emotional labels are embedded in their corresponding latent spaces to guide music generation. Third, we pretrain EmoMusicTV with the lead sheet continuation task to further improve its performance on conditional melody or harmony generation. Experimental results demonstrate that EmoMusicTV outperforms previous methods on three tasks, i.e., melody harmonization, melody generation given harmony, and lead sheet generation. Ablation studies verify the significant roles of emotional conditions and hierarchical latent variable structure on conditional music generation. Human listening shows that the lead sheets generated by EmoMusicTV are closer to the ground truth (GT) and perform slightly worse than the GT in conveying emotional polarity.

关键词： Symbolic music generation lead sheets emotional condition conditional variational autoencoder transformer

来源：评论

学校读者我要写书评

暂无评论

Multi-Label Zero-Shot Learning With Adversarial and variational Techniques

引用

IEEE ACCESS 2024年 12卷 94990-95006页

作者： Gull, Muqaddas Arif, Omar Natl Univ Sci & Technol NUST Sch Elect Engn & Comp Sci Islamabad 44000 Pakistan Amer Univ Sharjah Dept Comp Sci & Engn Sharjah U Arab Emirates

Multi-label zero-shot learning expands upon the traditional single-label zero-shot learning paradigm by addressing the challenge of accurately classifying images containing multiple unseen classes, which are not part of the training data. Current techniques rely on attention mechanisms to tackle the complexities of multi-label zero-shot learning (ZSL) and generalized zero-shot learning (GZSL). However, the generation of features, especially within the context of a generative approach, remains an unexplored area. In this paper, we propose a generative approach that leverages the capabilities of conditional variational autoencoder (CVAE) and conditional Generative Adversarial Network (CGAN) to enhance the quality of generative data for both multi-label ZSL and GZSL. Additionally, we introduce a novel "Regressor" as a supplementary tool to improve the reconstruction of visual features. This Regressor operates in conjunction with a "cycle-consistency loss" to ensure that the generated features preserve the key qualities of the original features even after undergoing transformations. To gauge the efficacy of our proposed approach, we conducted comprehensive experiments on two widely recognized benchmark datasets: NUS-WIDE and MS COCO. Our evaluation spanned both multi-label ZSL and GZSL scenarios. Notably, our approach yielded significant enhancements in mean Average Precision (mAP) for both datasets. Specifically, we observed a 0.2% increase in performance on the NUS-WIDE dataset and a notable 2.6% improvement on the MS COCO dataset in the context of Multi-label ZSL. The results clearly demonstrate that our generative approach outperforms existing methods on these widely-recognized datasets.

关键词： Semantics Visualization Generative adversarial networks Task analysis Data models Zero-shot learning Training Encoders conditional variational autoencoder conditional generative adversarial network generalized zero-shot learning regressor zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

Highly imbalanced fault diagnosis of turbine blade cracks via deep focal dynamically weighted conditional variational auto-encoder network

引用

ADVANCED ENGINEERING INFORMATICS 2024年第PartA期62卷

作者： Huang, Xin Zhang, Xiaodong Xiong, Yiwei Fan, Bochao Dai, Fei Xi An Jiao Tong Univ Sch Mech Engn Xian 710049 Peoples R China Xi An Jiao Tong Univ Key Lab Educ Minist Modern Design & Rotor Bearing Syst Xian 710049 Peoples R China

Turbine blade cracks pose a critical threat to the safety of aircraft in flight. However, data collected from turbine blades often exhibit substantial imbalances in real industry settings, posing challenges for effective blade crack diagnosis using intelligent models. To solve this problem, our study presents an intelligent fault diagnosis framework utilizing three-dimensional blade tip clearance (3D-BTC) signal under highly imbalanced samples, the so-called deep focal dynamically weighted conditional variational auto -encoder (DWCVAE-DFL), which can leverage the benefits of cost -sensitive learning methods and deep generative models to effectively tackles the diagnosis complexity of imbalanced class samples, thereby improving the blade crack diagnosis performance. Specifically, the novel deep focal loss (DFL) function is first designed to consider the varying sensitivity of imbalanced class samples in feature extraction and fault classification. The blade -rotor simulation rig is used for validating the effectiveness of our proposed method. Our experimental results indicate that the performance of our proposed method outperforms the latest methods when dealing with highly imbalanced samples.

关键词： Blade crack diagnosis Imbalanced class samples conditional variational autoencoder Deep focal loss Three-dimensional blade tip clearance

来源：评论

学校读者我要写书评

暂无评论

Deep learning-based surrogate models for spatial field solution reconstruction and uncertainty quantification in Structural Health Monitoring applications

引用

COMPUTERS & STRUCTURES 2024年 301卷

作者： Silionis, Nicholas E. Liangou, Theodora Anyfantis, Konstantinos N. Natl Tech Univ Athens Sch Naval Architecture & Marine Engn Ship Hull Struct Hlth Monitoring S H SHM Grp Heroon Polytech Ave Athens 15780 Greece

In recent years, increasingly complex computational models are being built to describe physical systems which has led to increased use of surrogate models to reduce computational cost. In problems related to Structural Health Monitoring (SHM), models capable of handling both high -dimensional data and quantifying uncertainty are required. In this work, our goal is to propose a conditional deep generative model as a surrogate aimed at such applications and high -dimensional stochastic structural simulations in general. To that end, a conditional variational autoencoder (CVAE) utilizing convolutional neural networks (CNNs) is employed to obtain reconstructions of spatially ordered structural response quantities for structural elements that are subjected to stochastic loading. Two numerical examples, inspired by potential SHM applications, are utilized to demonstrate the performance of the surrogate. The model is able to achieve high reconstruction accuracy compared to the reference Finite Element (FE) solutions, while at the same time successfully encoding the load uncertainty.

关键词： Surrogate modeling conditional variational autoencoder Structural Health Monitoring Probabilistic machine learning Deep generative models

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：