skip to main content
Language:
Search Limited to: Search Limited to: Resource type Show Results with: Show Results with: Search type Index

Improving AI System Awareness of Geoscience Knowledge: Symbiotic Integration of Physical Approaches and Deep Learning

Geophysical research letters, 2020-07, Vol.47 (13), p.n/a [Peer Reviewed Journal]

2020. The Authors. ;2020. This article is published under http://creativecommons.org/licenses/by-nc/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License. ;ISSN: 0094-8276 ;EISSN: 1944-8007 ;DOI: 10.1029/2020GL088229

Full text available

Citations Cited by
  • Title:
    Improving AI System Awareness of Geoscience Knowledge: Symbiotic Integration of Physical Approaches and Deep Learning
  • Author: Jiang, Shijie ; Zheng, Yi ; Solomatine, Dimitri
  • Subjects: Accuracy ; Artificial intelligence ; Catchment area ; Catchments ; Computer simulation ; Context ; Deep learning ; Earth ; Earth science ; Environmental studies ; Frameworks ; geosystem dynamics ; Hydrologic models ; Hydrology ; Instruments ; Integration ; Machine learning ; Model accuracy ; Modelling ; Physics ; predictions in ungauged basins ; Questions ; Representations ; Runoff ; Symbionts ; System dynamics ; Vision
  • Is Part Of: Geophysical research letters, 2020-07, Vol.47 (13), p.n/a
  • Description: Modeling dynamic geophysical phenomena is at the core of Earth and environmental studies. The geoscientific community relying mainly on physical representations may want to consider much deeper adoption of artificial intelligence (AI) instruments in the context of AI's global success and emergence of big Earth data. A new perspective of using hybrid physics‐AI approaches is a grand vision, but actualizing such approaches remains an open question in geoscience. This study develops a general approach to improving AI geoscientific awareness, wherein physical approaches such as temporal dynamic geoscientific models are included as special recurrent neural layers in a deep learning architecture. The illustrative case of runoff modeling across the conterminous United States demonstrates that the physics‐aware DL model has enhanced prediction accuracy, robust transferability, and good intelligence for inferring unobserved processes. This study represents a firm step toward realizing the vision of tackling Earth system challenges by physics‐AI integration. Plain Language Summary Artificial intelligence (AI) learns and makes inferences from experience and resembles the way untaught humans learn. If scientists can manage to teach AI the physical rules of the world, the “educated” AI may be more intelligent in deductions. However, how to implant elements of physical representations into an AI system effectively and directly remains an open question. This study proposes a general framework and applicable solutions to this challenge in the context of Earth science. The novel framework has a specially structured design for an AI system to “memorize” physical rules behind system dynamics (i.e., how a geosystem evolves with time). Following this framework, we developed a hydrology‐aware deep learning model to simulate/predict runoff in 569 catchments across the conterminous United States. The results show that after “learning” a hydrologic model, the AI system has enhanced prediction accuracy and good intelligence to deal with unfamiliar regions and infer unobserved processes. The potential of AI for in‐depth information mining, in return, fills the knowledge gap existing in physical approaches. The symbiotic integration of physical approaches and deep learning represents a promising solution to improve AI system awareness of geoscience knowledge. Key Points A novel approach to representing geosystem dynamics via a recurrent neural network within deep learning architectures is proposed The physics‐aware AI system exhibits robust transferability and good intelligence for inferring unobserved processes in runoff modeling The hydrology‐aware DL model can be an intelligent parameterization module for the encoded hydrologic model in cross‐region applications
  • Publisher: Washington: John Wiley & Sons, Inc
  • Language: English
  • Identifier: ISSN: 0094-8276
    EISSN: 1944-8007
    DOI: 10.1029/2020GL088229
  • Source: Wiley Blackwell AGU Digital Library
Back to results list
INSPIRE LIBRARY - TON DUC THANG UNIVERSITY (84-028) 37 755 057 Feedback
19 Nguyen Huu Tho St. Dist.7, HCM thuvien@tdtu.edu.vn

Copyright © 2017 INSPIRE LIBRARY - TON DUC THANG UNIVERSITY

Searching Remote Databases, Please Wait