We adapt graphical causal structure learning methods to apply to nonstationary time series data, specifically to processes that exhibit stochastic trends. We modify the likelihood component of the BIC score used by score-based search algorithms, such that it remains a consistent selection criterion for integrated or cointegrated processes. We use this modified score in conjunction with the SVAR-GFCI algorithm [15], which allows us to recover qualitative structural information about the underlying data-generating process even in the presence of latent (unmeasured) factors. We demonstrate our approach on both simulated and real macroeconomic data.
Download full-text PDF |
Source |
|---|---|
| http://www.ncbi.nlm.nih.gov/pmc/articles/PMC6890532 | PMC |
Publication Analysis
Top Keywords
Similar Publications
Machine learning-based forecasting of ground surface settlement induced by metro shield tunneling construction.
Sci Rep
December 2024
School of Civil Engineering and Architecture, Wuhan University of Technology, Wuhan, 430070, China.
Urban rail transit systems, represented by subways, have significantly alleviated the traffic pressure brought by urbanization and have addressed issues such as traffic congestion. However, as a commonly used construction method for subway tunnels, shield tunneling inevitably disturbs the surrounding soil, leading to uneven ground surface settlement, which can impact the safety of nearby buildings. Therefore, it is crucial to promptly obtain and predict the ground surface settlement induced by shield tunneling construction to enable safety warnings and evaluations.
View Article and Find Full Text PDFA two-level resolution neural network with enhanced interpretability for freeway traffic forecasting.
Sci Rep
December 2024
Department of Civil Engineering, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland.
Deep learning models are widely used for traffic forecasting on freeways due to their ability to learn complex temporal and spatial relationships. In particular, graph neural networks, which integrate graph theory into deep learning, have become popular for modeling traffic sensor networks. However, traditional graph convolutional networks (GCNs) face limitations in capturing long-range spatial correlations, which can hinder accurate long-term predictions.
View Article and Find Full Text PDFA machine learning based classifier for topological quantum materials.
Sci Rep
December 2024
Department of Electrical and Electronic Engineering, Bangladesh University of Engineering and Technology, Dhaka, 1205, Bangladesh.
Prediction and discovery of new materials with desired properties are at the forefront of quantum science and technology research. A major bottleneck in this field is the computational resources and time complexity related to finding new materials from ab initio calculations. In this work, an effective and robust deep learning-based model is proposed by incorporating persistent homology with graph neural network which offers an accuracy of and an F1 score of in classifying topological versus non-topological materials, outperforming the other state-of-the-art classifier models.
View Article and Find Full Text PDFElucidating governing factors of PFAS removal by polyamide membranes using machine learning and molecular simulations.
Nat Commun
December 2024
School of Civil & Environmental Engineering, Georgia Institute of Technology, Atlanta, GA, 30332, USA.
Per- and polyfluoroalkyl substances (PFASs) have recently garnered considerable concerns regarding their impacts on human and ecological health. Despite the important roles of polyamide membranes in remediating PFASs-contaminated water, the governing factors influencing PFAS transport across these membranes remain elusive. In this study, we investigate PFAS rejection by polyamide membranes using two machine learning (ML) models, namely XGBoost and multimodal transformer models.
View Article and Find Full Text PDFA lightweight and precision dual track 1D and 2D feature fusion convolutional network for machinery equipment fault diagnosis.
Sci Rep
December 2024
College of Mechanical and Electrical Engineering, Wenzhou University, Wenzhou, 325035, China.
Addressing the issues of a single-feature input channel structure, scarcity of training fault data, and insufficient feature learning capabilities in noisy environments for intelligent diagnostic models of mechanical equipment, we propose a method based on a one-dimensional and two-dimensional dual-channel feature information fusion convolutional neural network (1D_2DIFCNN). By constructing a one-dimensional and two-dimensiona dual-channel feature information fusion convolutional network and introducing a Convolutional Block Attention Mechanism, we utilize Random Overlapping Sampling Technique to process raw vibration signals. The model takes as inputs both one-dimensional data and two-dimensional Continuous Wavelet Transform images.
View Article and Find Full Text PDFWant AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!