Transparent AI: explainability of deep learning based load disaggregation

David Murray; Lina Stankovic; Vladimir Stankovic

doi:10.1145/3486611.3492410

Transparent AI: explainability of deep learning based load disaggregation

David Murray, Lina Stankovic, Vladimir Stankovic

Electronic And Electrical Engineering

Research output: Contribution to conference › Proceeding › peer-review

5 Citations (Scopus)

105 Downloads (Pure)

Abstract

The paper focuses on explaining the outputs of deep-learning based non-intrusive load monitoring (NILM). Explainability of NILM networks is needed for a range of stakeholders: (i) technology developers to understand why a model is under/over predicting energy usage, missing appliances or false positives, (ii) businesses offering energy advice based on NILM as part of a broader energy home management recommender system, and (iii) end-users who need to understand the outcomes of the NILM inference.

Original language	English
Pages	268-271
Number of pages	4
DOIs	https://doi.org/10.1145/3486611.3492410
Publication status	Published - 17 Nov 2021
Event	The 1st ACM SIGEnergy Workshop of Fair, Accountable, Transparent, and Ethical AI for Smart Environments and Energy Systems - Coimbra, Portugal Duration: 17 Nov 2021 → 18 Nov 2021 https://fatesys.github.io/2021/

Workshop

Workshop	The 1st ACM SIGEnergy Workshop of Fair, Accountable, Transparent, and Ethical AI for Smart Environments and Energy Systems
Abbreviated title	FATEsys 2021
Country/Territory	Portugal
City	Coimbra
Period	17/11/21 → 18/11/21
Internet address	https://fatesys.github.io/2021/

Keywords

datasets
neural networks
reliability
validation

UN SDGs

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

Access to Document

10.1145/3486611.3492410

Murray-etal-FATEsys-2021-Transparent-AI-explainability-of-deep-learning-based-load-disaggregationAccepted author manuscript, 403 KB

Cite this

@conference{337a3703ce4143a79534763ee069be7a,

title = "Transparent AI: explainability of deep learning based load disaggregation",

abstract = "The paper focuses on explaining the outputs of deep-learning based non-intrusive load monitoring (NILM). Explainability of NILM networks is needed for a range of stakeholders: (i) technology developers to understand why a model is under/over predicting energy usage, missing appliances or false positives, (ii) businesses offering energy advice based on NILM as part of a broader energy home management recommender system, and (iii) end-users who need to understand the outcomes of the NILM inference.",

keywords = "datasets, neural networks, reliability, validation",

author = "David Murray and Lina Stankovic and Vladimir Stankovic",

year = "2021",

month = nov,

day = "17",

doi = "10.1145/3486611.3492410",

language = "English",

pages = "268--271",

note = "The 1st ACM SIGEnergy Workshop of Fair, Accountable, Transparent, and Ethical AI for Smart Environments and Energy Systems, FATEsys 2021 ; Conference date: 17-11-2021 Through 18-11-2021",

url = "https://fatesys.github.io/2021/",

}

TY - CONF

T1 - Transparent AI

T2 - The 1st ACM SIGEnergy Workshop of Fair, Accountable, Transparent, and Ethical AI for Smart Environments and Energy Systems

AU - Murray, David

AU - Stankovic, Lina

AU - Stankovic, Vladimir

PY - 2021/11/17

Y1 - 2021/11/17

N2 - The paper focuses on explaining the outputs of deep-learning based non-intrusive load monitoring (NILM). Explainability of NILM networks is needed for a range of stakeholders: (i) technology developers to understand why a model is under/over predicting energy usage, missing appliances or false positives, (ii) businesses offering energy advice based on NILM as part of a broader energy home management recommender system, and (iii) end-users who need to understand the outcomes of the NILM inference.

AB - The paper focuses on explaining the outputs of deep-learning based non-intrusive load monitoring (NILM). Explainability of NILM networks is needed for a range of stakeholders: (i) technology developers to understand why a model is under/over predicting energy usage, missing appliances or false positives, (ii) businesses offering energy advice based on NILM as part of a broader energy home management recommender system, and (iii) end-users who need to understand the outcomes of the NILM inference.

KW - datasets

KW - neural networks

KW - reliability

KW - validation

UR - https://fatesys.github.io/2021/

U2 - 10.1145/3486611.3492410

DO - 10.1145/3486611.3492410

M3 - Proceeding

SP - 268

EP - 271

Y2 - 17 November 2021 through 18 November 2021

ER -

Transparent AI: explainability of deep learning based load disaggregation

Abstract

Workshop

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

building GrEener and more sustainable soCieties by filling the Knowledge gap in social science and engineering responsible artificial intelligence co-creatiOn (GECKO) MSCA-ITN-2020

SENSIBLE: SENSors and Intelligence in BuiLt Environment (SENSIBLE) MSCA RISE

REFIT: Electrical Load Measurements (Cleaned)

Cite this

Transparent AI: explainability of deep learning based load disaggregation

Abstract

Workshop

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Projects

building GrEener and more sustainable soCieties by filling the Knowledge gap in social science and engineering responsible artificial intelligence co-creatiOn (GECKO) MSCA-ITN-2020

SENSIBLE: SENSors and Intelligence in BuiLt Environment (SENSIBLE) MSCA RISE

Datasets

REFIT: Electrical Load Measurements (Cleaned)

Cite this