Filling Gaps in Micro-meteorological Data

Antoine Richard; Lior Fine; Offer Rozenstein; Josef Tanny; Matthieu Geist; Cedric Pradalier

doi:10.1007/978-3-030-67670-4_7

Filling Gaps in Micro-meteorological Data

Antoine Richard, Lior Fine, Offer Rozenstein, Josef Tanny, Matthieu Geist, Cedric Pradalier

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

3 Scopus citations

Abstract

Filling large data-gaps in Micro-Meteorological data has mostly been done using interpolation techniques based on a marginal distribution sampling. Those methods work well but need a large horizon of the previous events to achieve good results since they do not model the system but only rely on previously encountered iterations. In this paper, we propose to use multi-head deep attention networks to fill gaps in Micro-Meteorological Data. This methodology couples large-scale information extraction with modeling capabilities that cannot be achieved by interpolation-like techniques. Unlike Bidirectional RNNs, our architecture is not recurrent, it is simple to tune and our data efficiency is higher. We apply our architecture to real-life data and clearly show its applicability in agriculture, furthermore, we show that it could be used to solve related problems such as filling gaps in cyclic-multivariate-time-series.

Original language	English
Title of host publication	Machine Learning and Knowledge Discovery in Databases. Applied Data Science and Demo Track - European Conference, ECML PKDD 2020, Proceedings
Editors	Yuxiao Dong, Dunja Mladenic, Craig Saunders
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	101-117
Number of pages	17
ISBN (Print)	9783030676698
DOIs	https://doi.org/10.1007/978-3-030-67670-4_7
State	Published - 1 Jan 2021
Externally published	Yes
Event	European Conference on Machine Learning and Knowledge Discovery in Databases, ECML PKDD 2020 - Virtual, Online Duration: 14 Sep 2020 → 18 Sep 2020

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12461 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	European Conference on Machine Learning and Knowledge Discovery in Databases, ECML PKDD 2020
City	Virtual, Online
Period	14/09/20 → 18/09/20

Keywords

Attention-models
Evapo-transpiration
Gap-filling

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-030-67670-4_7

Cite this

Richard, A., Fine, L., Rozenstein, O., Tanny, J., Geist, M., & Pradalier, C. (2021). Filling Gaps in Micro-meteorological Data. In Y. Dong, D. Mladenic, & C. Saunders (Eds.), Machine Learning and Knowledge Discovery in Databases. Applied Data Science and Demo Track - European Conference, ECML PKDD 2020, Proceedings (pp. 101-117). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12461 LNAI). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-67670-4_7

Richard, Antoine ; Fine, Lior ; Rozenstein, Offer et al. / Filling Gaps in Micro-meteorological Data. Machine Learning and Knowledge Discovery in Databases. Applied Data Science and Demo Track - European Conference, ECML PKDD 2020, Proceedings. editor / Yuxiao Dong ; Dunja Mladenic ; Craig Saunders. Springer Science and Business Media Deutschland GmbH, 2021. pp. 101-117 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{8d30262d495d486e9b7348cb61c81bf3,

title = "Filling Gaps in Micro-meteorological Data",

abstract = "Filling large data-gaps in Micro-Meteorological data has mostly been done using interpolation techniques based on a marginal distribution sampling. Those methods work well but need a large horizon of the previous events to achieve good results since they do not model the system but only rely on previously encountered iterations. In this paper, we propose to use multi-head deep attention networks to fill gaps in Micro-Meteorological Data. This methodology couples large-scale information extraction with modeling capabilities that cannot be achieved by interpolation-like techniques. Unlike Bidirectional RNNs, our architecture is not recurrent, it is simple to tune and our data efficiency is higher. We apply our architecture to real-life data and clearly show its applicability in agriculture, furthermore, we show that it could be used to solve related problems such as filling gaps in cyclic-multivariate-time-series.",

keywords = "Attention-models, Evapo-transpiration, Gap-filling",

author = "Antoine Richard and Lior Fine and Offer Rozenstein and Josef Tanny and Matthieu Geist and Cedric Pradalier",

note = "Publisher Copyright: {\textcopyright} 2021, Springer Nature Switzerland AG.; European Conference on Machine Learning and Knowledge Discovery in Databases, ECML PKDD 2020 ; Conference date: 14-09-2020 Through 18-09-2020",

year = "2021",

month = jan,

day = "1",

doi = "10.1007/978-3-030-67670-4_7",

language = "English",

isbn = "9783030676698",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "101--117",

editor = "Yuxiao Dong and Dunja Mladenic and Craig Saunders",

booktitle = "Machine Learning and Knowledge Discovery in Databases. Applied Data Science and Demo Track - European Conference, ECML PKDD 2020, Proceedings",

address = "Germany",

}

Richard, A, Fine, L, Rozenstein, O, Tanny, J, Geist, M & Pradalier, C 2021, Filling Gaps in Micro-meteorological Data. in Y Dong, D Mladenic & C Saunders (eds), Machine Learning and Knowledge Discovery in Databases. Applied Data Science and Demo Track - European Conference, ECML PKDD 2020, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12461 LNAI, Springer Science and Business Media Deutschland GmbH, pp. 101-117, European Conference on Machine Learning and Knowledge Discovery in Databases, ECML PKDD 2020, Virtual, Online, 14/09/20. https://doi.org/10.1007/978-3-030-67670-4_7

Filling Gaps in Micro-meteorological Data. / Richard, Antoine; Fine, Lior; Rozenstein, Offer et al.
Machine Learning and Knowledge Discovery in Databases. Applied Data Science and Demo Track - European Conference, ECML PKDD 2020, Proceedings. ed. / Yuxiao Dong; Dunja Mladenic; Craig Saunders. Springer Science and Business Media Deutschland GmbH, 2021. p. 101-117 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12461 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Filling Gaps in Micro-meteorological Data

AU - Richard, Antoine

AU - Fine, Lior

AU - Rozenstein, Offer

AU - Tanny, Josef

AU - Geist, Matthieu

AU - Pradalier, Cedric

PY - 2021/1/1

Y1 - 2021/1/1

N2 - Filling large data-gaps in Micro-Meteorological data has mostly been done using interpolation techniques based on a marginal distribution sampling. Those methods work well but need a large horizon of the previous events to achieve good results since they do not model the system but only rely on previously encountered iterations. In this paper, we propose to use multi-head deep attention networks to fill gaps in Micro-Meteorological Data. This methodology couples large-scale information extraction with modeling capabilities that cannot be achieved by interpolation-like techniques. Unlike Bidirectional RNNs, our architecture is not recurrent, it is simple to tune and our data efficiency is higher. We apply our architecture to real-life data and clearly show its applicability in agriculture, furthermore, we show that it could be used to solve related problems such as filling gaps in cyclic-multivariate-time-series.

AB - Filling large data-gaps in Micro-Meteorological data has mostly been done using interpolation techniques based on a marginal distribution sampling. Those methods work well but need a large horizon of the previous events to achieve good results since they do not model the system but only rely on previously encountered iterations. In this paper, we propose to use multi-head deep attention networks to fill gaps in Micro-Meteorological Data. This methodology couples large-scale information extraction with modeling capabilities that cannot be achieved by interpolation-like techniques. Unlike Bidirectional RNNs, our architecture is not recurrent, it is simple to tune and our data efficiency is higher. We apply our architecture to real-life data and clearly show its applicability in agriculture, furthermore, we show that it could be used to solve related problems such as filling gaps in cyclic-multivariate-time-series.

KW - Attention-models

KW - Evapo-transpiration

KW - Gap-filling

UR - http://www.scopus.com/inward/record.url?scp=85103292544&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-67670-4_7

DO - 10.1007/978-3-030-67670-4_7

M3 - Conference contribution

AN - SCOPUS:85103292544

SN - 9783030676698

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 101

EP - 117

BT - Machine Learning and Knowledge Discovery in Databases. Applied Data Science and Demo Track - European Conference, ECML PKDD 2020, Proceedings

A2 - Dong, Yuxiao

A2 - Mladenic, Dunja

A2 - Saunders, Craig

PB - Springer Science and Business Media Deutschland GmbH

T2 - European Conference on Machine Learning and Knowledge Discovery in Databases, ECML PKDD 2020

Y2 - 14 September 2020 through 18 September 2020

ER -

Richard A, Fine L, Rozenstein O, Tanny J, Geist M, Pradalier C. Filling Gaps in Micro-meteorological Data. In Dong Y, Mladenic D, Saunders C, editors, Machine Learning and Knowledge Discovery in Databases. Applied Data Science and Demo Track - European Conference, ECML PKDD 2020, Proceedings. Springer Science and Business Media Deutschland GmbH. 2021. p. 101-117. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-67670-4_7

Filling Gaps in Micro-meteorological Data

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this