Van der Pol-informed Neural Networks for Multi-step-ahead Forecasting of Extreme Climatic Events

Deep learning has produced excellent results in several applied domains including computer vision, natural language processing, speech recognition, etc. Physics-informed neural networks (PINN) are a new family of deep learning models that combine prior knowledge of physics in the form of high-level...

Full description

Saved in:
Bibliographic Details
Main Author: Dutta, Anurag (author)
Other Authors: Panja, Madhurima (author), Kumar, Uttam (author), Hens, Chittaranjan (author), Chakraborty, Tanujit (author)
Published: 2023
Online Access:https://depot.sorbonne.ae/handle/20.500.12458/1462
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1857415063125622784
author Dutta, Anurag
author2 Panja, Madhurima
Kumar, Uttam
Hens, Chittaranjan
Chakraborty, Tanujit
author2_role author
author
author
author
author_facet Dutta, Anurag
Panja, Madhurima
Kumar, Uttam
Hens, Chittaranjan
Chakraborty, Tanujit
author_role author
dc.creator.none.fl_str_mv Dutta, Anurag
Panja, Madhurima
Kumar, Uttam
Hens, Chittaranjan
Chakraborty, Tanujit
dc.date.none.fl_str_mv 2023
2024-01-30T09:48:19Z
2024-01-30T09:48:19Z
dc.format.none.fl_str_mv application/pdf
dc.identifier.none.fl_str_mv https://depot.sorbonne.ae/handle/20.500.12458/1462
dc.language.none.fl_str_mv en
dc.relation.none.fl_str_mv 37th Conference on Neural Information Processing Systems
dc.title.none.fl_str_mv Van der Pol-informed Neural Networks for Multi-step-ahead Forecasting of Extreme Climatic Events
dc.type.none.fl_str_mv Controlled Vocabulary for Resource Type Genres::text::conference object::conference proceedings
description Deep learning has produced excellent results in several applied domains including computer vision, natural language processing, speech recognition, etc. Physics-informed neural networks (PINN) are a new family of deep learning models that combine prior knowledge of physics in the form of high-level abstraction of natural phenomena with data-driven neural networks. PINN has emerged as a flourishing area of scientific computing to deal with the challenges of shortage of training data, enhancing physical plausibility, and specifically aiming to solve complex differential equations. However, building PINNs for modeling and forecasting the dynamics of extreme climatic events of geophysical systems remains an open scientific problem. This study proposes Van der Pol-informed Neural Networks (VPINN), a physics-informed differential learning approach, for modeling extreme nonlinear dynamical systems such as climatic events, exploiting the physical differentials as the physics-derived loss function. Our proposal is compared to state-of-the-art time series forecasting models, showing superior performance.The codes and dataset used for the experiments are made available at https: //github.com/mad-stat/VPINN.
id sorbonner_61a166dcaac656121d1ad9c46a2cfde8
language_invalid_str_mv en
network_acronym_str sorbonner
network_name_str Sorbonne University Abu Dhabi repository
oai_identifier_str oai:depot.sorbonne.ae:20.500.12458/1462
publishDate 2023
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
spelling Van der Pol-informed Neural Networks for Multi-step-ahead Forecasting of Extreme Climatic EventsDutta, AnuragPanja, MadhurimaKumar, UttamHens, ChittaranjanChakraborty, TanujitDeep learning has produced excellent results in several applied domains including computer vision, natural language processing, speech recognition, etc. Physics-informed neural networks (PINN) are a new family of deep learning models that combine prior knowledge of physics in the form of high-level abstraction of natural phenomena with data-driven neural networks. PINN has emerged as a flourishing area of scientific computing to deal with the challenges of shortage of training data, enhancing physical plausibility, and specifically aiming to solve complex differential equations. However, building PINNs for modeling and forecasting the dynamics of extreme climatic events of geophysical systems remains an open scientific problem. This study proposes Van der Pol-informed Neural Networks (VPINN), a physics-informed differential learning approach, for modeling extreme nonlinear dynamical systems such as climatic events, exploiting the physical differentials as the physics-derived loss function. Our proposal is compared to state-of-the-art time series forecasting models, showing superior performance.The codes and dataset used for the experiments are made available at https: //github.com/mad-stat/VPINN.2024-01-30T09:48:19Z2024-01-30T09:48:19Z2023Controlled Vocabulary for Resource Type Genres::text::conference object::conference proceedingsapplication/pdfhttps://depot.sorbonne.ae/handle/20.500.12458/1462en37th Conference on Neural Information Processing Systemsoai:depot.sorbonne.ae:20.500.12458/14622024-03-10T07:10:39Z
spellingShingle Van der Pol-informed Neural Networks for Multi-step-ahead Forecasting of Extreme Climatic Events
Dutta, Anurag
title Van der Pol-informed Neural Networks for Multi-step-ahead Forecasting of Extreme Climatic Events
title_full Van der Pol-informed Neural Networks for Multi-step-ahead Forecasting of Extreme Climatic Events
title_fullStr Van der Pol-informed Neural Networks for Multi-step-ahead Forecasting of Extreme Climatic Events
title_full_unstemmed Van der Pol-informed Neural Networks for Multi-step-ahead Forecasting of Extreme Climatic Events
title_short Van der Pol-informed Neural Networks for Multi-step-ahead Forecasting of Extreme Climatic Events
title_sort Van der Pol-informed Neural Networks for Multi-step-ahead Forecasting of Extreme Climatic Events
url https://depot.sorbonne.ae/handle/20.500.12458/1462