Related references
Note: Only part of the references are listed.
Article
Engineering, Mechanical
Xiaoming Lei et al.
Summary: This study proposes a residual attention network (RAN) that employs the attention mechanism and residual learning to improve classification efficiency and accuracy, aiming to address the impact of abnormal monitoring data on structural assessment. The hourly segmented measured data is transformed into matrix form through mutual information correlation analysis for training a deep learning model. The proposed RAN model shows excellent classification performance in identifying most anomaly data types in the test dataset and demonstrates good generalization performance with another cable-stayed bridge dataset, outperforming existing preprocessing and deep learning models in multi-classification and classification accuracy.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2023)
Article
Computer Science, Artificial Intelligence
Sentao Chen et al.
Summary: In this work, we propose a solution to the problem of domain generalization for classification, which involves learning a classification model on a set of source domains and generalizing it to a target domain. Our approach aligns a joint distribution and a product distribution using a neural transformation and minimizes the Relative Chi-Square (RCS) divergence between the two distributions to learn the transformation. We demonstrate the effectiveness of our solution through comparisons with state-of-the-art methods on various image classification datasets.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Artificial Intelligence
Mattia Segu et al.
Summary: Domain generalization aims to train machine learning models that can perform robustly across different domains. This study proposes a new approach that explicitly trains domain-dependent representations and maps domains in a shared latent space, achieving better generalization performance.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Artificial Intelligence
Sijie Hu et al.
Summary: In this paper, a novel multi-modal-based Unsupervised Domain Adaptation (UDA) method named MMADT is proposed for semantic segmentation. It utilizes both RGB and depth images and introduces a Depth Fusion Block (DFB) and Depth Adversarial Training (DAT) to address the depth discrepancy between the source and target domain. Besides, a self-supervised multi-modal depth estimation assistant network named Geo-Assistant (GA) is proposed to align the feature space of RGB and depth. Experimental results show significant performance improvement in multiple synthetic to real adaptation benchmarks.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Interdisciplinary Applications
Luca Trani et al.
Summary: Recent developments in instrumentation and methodology have driven advances in solid Earth sciences, allowing data centres to acquire larger datasets of higher quality. Artificial Intelligence methods, such as Convolutional Neural Networks (CNN), offer opportunities to analyze this growing information potential. This paper investigates the use of CNN for seismo-acoustic event classification in the Netherlands, proposing a specialized CNN architecture that achieves high performance in source detection. The study also focuses on explainability aspects and introduces a method to gain insights into the behavior of deep learning classification systems, providing a usable tool for domain experts and promoting the operational implementation of AI-based methods.
COMPUTERS & GEOSCIENCES
(2022)
Article
Geochemistry & Geophysics
Jinsheng Jiang et al.
Summary: Petroleum geophysical exploration based on seismic data has been influenced by deep learning technology. The improved CAE method for simultaneous reconstruction and denoising of seismic data shows effectiveness in feature extraction and noise suppression.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Computer Science, Artificial Intelligence
Xianchao Zhang et al.
Summary: The study introduces a method called Deep Adversarial Anomaly Detection (DAAD) to overcome the limitations of existing anomaly detection methods. By modeling the marginal distributions of normal data and utilizing adversarial training, the performance of anomaly detection is improved. Reliable results are obtained through a majority voting strategy.
PATTERN RECOGNITION
(2022)
Article
Computer Science, Information Systems
Yunyun Wang et al.
Summary: This paper introduces a self-corrected unsupervised domain adaptation method called SCUDA, which uses a probabilistic label corrector to learn and correct target labels directly. Unlike traditional UDA methods, SCUDA attempts to learn the target prediction end to end.
FRONTIERS OF COMPUTER SCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Diogo Pernes et al.
Summary: This work presents a novel framework that addresses the problems of adjusting the mixture distribution weights in multi-source domain adaptation and ensuring low error on the target domain. By using a mildly optimistic objective function and consistency regularization on the target samples, it surpasses the current state of the art.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Ignacio Peis et al.
Summary: The study introduces a novel deep generative model that captures global dependencies in a fully unsupervised manner by combining a mixture model in the local space and a global Gaussian latent variable. The model is able to capture interpretable disentangled representations, perform domain alignment, and discriminate between groups with non-trivial structures.
PATTERN RECOGNITION
(2022)
Article
Multidisciplinary Sciences
Irene Ndunge Muisyo et al.
Summary: This paper investigates the utilization of a STATCOM to enhance the LVRT capability of a 9 MW DFIG based Wind Power Plant and compares the performance of STATCOM tuned using different algorithms. The results show that the STATCOM tuned with the hybrid WCA-PSO algorithm performs the best, meeting the LVRT capability requirements of the Danish power system.
Article
Construction & Building Technology
Can Jin et al.
Summary: This study proposes a method to accurately classify the morphology of aggregates and investigates the influence of different morphologies on the internal structure of asphalt mixtures. By evaluating the similarities of aggregates and using clustering methods, the aggregates are classified and morphology labels are constructed. This method provides insights for optimizing the internal structure through the configuration of aggregate morphologies.
CONSTRUCTION AND BUILDING MATERIALS
(2022)
Article
Engineering, Geological
Umer Waqas et al.
Summary: This study investigates the strength behavior of thermally damaged rocks under dynamic cyclic loading, and utilizes advanced data modeling techniques to establish a predictive relationship between SIF and dynamic properties. Comparing different classification algorithms, a hybrid meta-classifier (MC) is found to perform the best in rock classification based on SIF.
INTERNATIONAL JOURNAL OF ROCK MECHANICS AND MINING SCIENCES
(2022)
Article
Energy & Fuels
Ting Xu et al.
Summary: This study investigates the domain generalization problem in lithology identification, proposing a contrastive domain discrepancy based adversarial autoencoder (CDD-AAE) framework to ensure prediction accuracy on a new unseen well.
JOURNAL OF NATURAL GAS SCIENCE AND ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Carlos Santiago et al.
Summary: The proposed strategy LOW for training DL models by learning optimal sample weights improves model performance by better utilizing training data, leading to faster convergence and better generalization, especially on imbalanced datasets.
PATTERN RECOGNITION
(2021)
Article
Computer Science, Information Systems
Jialie Shen et al.
Summary: Recent years have seen rapid development of deep neural networks (DNN) and increasing interest in adversarial example attacks. Researchers have proposed an ensemble-based approach to enhance the robustness and reliability of DNN models, and introduced the BBAS scheme for diverse adversarial example generation.
INFORMATION SCIENCES
(2021)
Article
Energy & Fuels
Wei Li et al.
Summary: The study proposes a new method to improve seismic interference and enhance the prediction accuracy of sandstone thickness. By combining multiple seismic attributes and implementing supervised machine learning using support vector regression, the effectiveness of the method was demonstrated in an oilfield in China.
JOURNAL OF PETROLEUM SCIENCE AND ENGINEERING
(2021)
Article
Computer Science, Artificial Intelligence
Qiangguo Jin et al.
Summary: This study proposes a prior knowledge driven domain adaptation and dual-domain enhanced self-correction learning scheme for segmentation of COVID-19 infection on CT images. The DASC-Net model consistently outperforms state-of-the-art segmentation, domain shift, and coronavirus infection segmentation methods, enriching the theory of domain adaptation and self-correction learning in medical imaging.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Information Systems
Andrea Rosales Sanabria et al.
Summary: Sensor-based human activity recognition (HAR) plays a significant role in various fields such as smart city, smart home, and personal healthcare. The shift-GAN proposed in this article integrates Bi-GAN and KMM to achieve robust feature transfer between two different domains. Experimental results demonstrate that shift-GAN outperforms over 10 existing domain adaptation techniques, learns intrinsic feature mappings independent of activities between two domains, is robust to sensor noise, and less sensitive to training data.
Article
Engineering, Biomedical
S. Thomas George et al.
BIOCYBERNETICS AND BIOMEDICAL ENGINEERING
(2020)
Article
Engineering, Electrical & Electronic
M. A. Amirabadi et al.
PHYSICAL COMMUNICATION
(2020)