AI explainability framework for environmental management research

Mehrdad Arashpour

doi:10.1016/j.jenvman.2023.118149

AI explainability framework for environmental management research

Mehrdad Arashpour

Civil Engineering

Research output: Contribution to journal › Article › Research › peer-review

25 Citations (Scopus)

Abstract

Deep learning networks powered by AI are essential predictive tools relying on image data availability and processing hardware advancements. However, little attention has been paid to explainable AI (XAI) in application fields, including environmental management. This study develops an explainability framework with a triadic structure to focus on input, AI model and output. The framework provides three main contributions. (1) A context-based augmentation of input data to maximize generalizability and minimize overfitting. (2) A direct monitoring of AI model layers and parameters to use leaner (lighter) networks suitable for edge device deployment, (3) An output explanation procedure focusing on interpretability and robustness of predictive decisions by AI networks. These contributions significantly advance state of the art in XAI for environmental management research, offering implications for improved understanding and utilization of AI networks in this field.

Original language	English
Article number	118149
Number of pages	7
Journal	Journal of Environmental Management
Volume	342
DOIs	https://doi.org/10.1016/j.jenvman.2023.118149
Publication status	Published - 15 Sept 2023

Keywords

Environmental crisis
Environmental management research
Explainable AI (XAI)
Management and valorization of solid waste
Multimodal and generative pre-trained transformers
Responsible and fair artificial intelligence
Vision-language deep learning models

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1016/j.jenvman.2023.118149Licence: CC BY

Cite this

@article{9336daf6cf3c4c89bfe2e2cd36592225,

title = "AI explainability framework for environmental management research",

abstract = "Deep learning networks powered by AI are essential predictive tools relying on image data availability and processing hardware advancements. However, little attention has been paid to explainable AI (XAI) in application fields, including environmental management. This study develops an explainability framework with a triadic structure to focus on input, AI model and output. The framework provides three main contributions. (1) A context-based augmentation of input data to maximize generalizability and minimize overfitting. (2) A direct monitoring of AI model layers and parameters to use leaner (lighter) networks suitable for edge device deployment, (3) An output explanation procedure focusing on interpretability and robustness of predictive decisions by AI networks. These contributions significantly advance state of the art in XAI for environmental management research, offering implications for improved understanding and utilization of AI networks in this field.",

keywords = "Environmental crisis, Environmental management research, Explainable AI (XAI), Management and valorization of solid waste, Multimodal and generative pre-trained transformers, Responsible and fair artificial intelligence, Vision-language deep learning models",

author = "Mehrdad Arashpour",

note = "Funding Information: The contributions and comments of ASCII Lab ( https://www.monash.edu/ascii ) members on this manuscript are greatly appreciated. This work is part of a research project (Project No IH18.04.1 ) funded by the Australian Research Council (ARC) Industrial Transformation Research Hub (ITRH) Scheme (Project ID: IH180100010 ). Publisher Copyright: {\textcopyright} 2023 The Author",

year = "2023",

month = sep,

day = "15",

doi = "10.1016/j.jenvman.2023.118149",

language = "English",

volume = "342",

journal = "Journal of Environmental Management",

issn = "0301-4797",

publisher = "Elsevier BV",

}

TY - JOUR

T1 - AI explainability framework for environmental management research

AU - Arashpour, Mehrdad

N1 - Funding Information: The contributions and comments of ASCII Lab ( https://www.monash.edu/ascii ) members on this manuscript are greatly appreciated. This work is part of a research project (Project No IH18.04.1 ) funded by the Australian Research Council (ARC) Industrial Transformation Research Hub (ITRH) Scheme (Project ID: IH180100010 ). Publisher Copyright: © 2023 The Author

PY - 2023/9/15

Y1 - 2023/9/15

N2 - Deep learning networks powered by AI are essential predictive tools relying on image data availability and processing hardware advancements. However, little attention has been paid to explainable AI (XAI) in application fields, including environmental management. This study develops an explainability framework with a triadic structure to focus on input, AI model and output. The framework provides three main contributions. (1) A context-based augmentation of input data to maximize generalizability and minimize overfitting. (2) A direct monitoring of AI model layers and parameters to use leaner (lighter) networks suitable for edge device deployment, (3) An output explanation procedure focusing on interpretability and robustness of predictive decisions by AI networks. These contributions significantly advance state of the art in XAI for environmental management research, offering implications for improved understanding and utilization of AI networks in this field.

AB - Deep learning networks powered by AI are essential predictive tools relying on image data availability and processing hardware advancements. However, little attention has been paid to explainable AI (XAI) in application fields, including environmental management. This study develops an explainability framework with a triadic structure to focus on input, AI model and output. The framework provides three main contributions. (1) A context-based augmentation of input data to maximize generalizability and minimize overfitting. (2) A direct monitoring of AI model layers and parameters to use leaner (lighter) networks suitable for edge device deployment, (3) An output explanation procedure focusing on interpretability and robustness of predictive decisions by AI networks. These contributions significantly advance state of the art in XAI for environmental management research, offering implications for improved understanding and utilization of AI networks in this field.

KW - Environmental crisis

KW - Environmental management research

KW - Explainable AI (XAI)

KW - Management and valorization of solid waste

KW - Multimodal and generative pre-trained transformers

KW - Responsible and fair artificial intelligence

KW - Vision-language deep learning models

UR - http://www.scopus.com/inward/record.url?scp=85159040416&partnerID=8YFLogxK

U2 - 10.1016/j.jenvman.2023.118149

DO - 10.1016/j.jenvman.2023.118149

M3 - Article

AN - SCOPUS:85159040416

SN - 0301-4797

VL - 342

JO - Journal of Environmental Management

JF - Journal of Environmental Management

M1 - 118149

ER -

AI explainability framework for environmental management research

Abstract

Keywords

UN SDGs

Access to Document

Other files and links

SPARC: ARC Research Hub for Smart Next Generation Transport Pavements

Cite this

AI explainability framework for environmental management research

Abstract

Keywords

UN SDGs

Access to Document

Other files and links

Projects

SPARC: ARC Research Hub for Smart Next Generation Transport Pavements

Cite this