An LSTM-Based Method for Detection and Classification of Sensor Anomalies

Alexander Verner; Sumitra Mukherjee

doi:10.1145/3409073.3409089

An LSTM-Based Method for Detection and Classification of Sensor Anomalies

Alexander Verner
, Sumitra Mukherjee

Nova Southeastern University

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

Abstract

Most existing machine learning (ML) based solutions for anomaly detection in sensory data rely on carefully hand-crafted features. This approach has a fundamental limitation since it is often application-specific and requires considerable human effort from domain experts. Deep learning models have been demonstrated to have the ability to abstract relevant high-level features from raw data. Long short-Term memory (LSTM) recurrent neural networks have proven effective in complex time-series prediction problems. In this paper, we propose an LSTM-based method for anomaly detection in sensory data. We systematically investigate its effectiveness on raw time-series of real medical sensors measurements and show that it achieves the same level of performance as traditional ML models operating on carefully designed feature vectors. The proposed method achieved micro, macro, and weighted precision, recall, and F1-score of over 0.99.

Original language	English
Title of host publication	ICMLT 2020 - Proceedings of 2020 5th International Conference on Machine Learning Technologies
Publisher	Association for Computing Machinery (ACM)
Pages	39-45
Number of pages	7
ISBN (Electronic)	9781450377645
ISBN (Print)	9781450377645
DOIs	https://doi.org/10.1145/3409073.3409089
State	Published - Jun 19 2020
Event	5th International Conference on Machine Learning Technologies, ICMLT 2020 - Beijing, Online, China Duration: Jun 19 2020 → Jun 21 2020

Publication series

Name	Proceedings of the 2020 5th International Conference on Machine Learning Technologies

Conference

Conference	5th International Conference on Machine Learning Technologies, ICMLT 2020
Country/Territory	China
City	Beijing, Online
Period	6/19/20 → 6/21/20

Bibliographical note

Publisher Copyright:
© 2020 ACM.

ASJC Scopus Subject Areas

Software
Human-Computer Interaction
Computer Vision and Pattern Recognition
Computer Networks and Communications

Keywords

Anomaly detection
deep learning
LSTM
machine learning
sensors validation
time-series analysis

Access to Document

10.1145/3409073.3409089

Cite this

Verner, A., & Mukherjee, S. (2020). An LSTM-Based Method for Detection and Classification of Sensor Anomalies. In ICMLT 2020 - Proceedings of 2020 5th International Conference on Machine Learning Technologies (pp. 39-45). (Proceedings of the 2020 5th International Conference on Machine Learning Technologies). Association for Computing Machinery (ACM). https://doi.org/10.1145/3409073.3409089

An LSTM-Based Method for Detection and Classification of Sensor Anomalies. / Verner, Alexander; Mukherjee, Sumitra.
ICMLT 2020 - Proceedings of 2020 5th International Conference on Machine Learning Technologies. Association for Computing Machinery (ACM), 2020. p. 39-45 (Proceedings of the 2020 5th International Conference on Machine Learning Technologies).

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

Verner, A & Mukherjee, S 2020, An LSTM-Based Method for Detection and Classification of Sensor Anomalies. in ICMLT 2020 - Proceedings of 2020 5th International Conference on Machine Learning Technologies. Proceedings of the 2020 5th International Conference on Machine Learning Technologies, Association for Computing Machinery (ACM), pp. 39-45, 5th International Conference on Machine Learning Technologies, ICMLT 2020, Beijing, Online, China, 6/19/20. https://doi.org/10.1145/3409073.3409089

Verner A, Mukherjee S. An LSTM-Based Method for Detection and Classification of Sensor Anomalies. In ICMLT 2020 - Proceedings of 2020 5th International Conference on Machine Learning Technologies. Association for Computing Machinery (ACM). 2020. p. 39-45. (Proceedings of the 2020 5th International Conference on Machine Learning Technologies). doi: 10.1145/3409073.3409089

@inproceedings{ee24bc66810743a0a11e558eea71a6fa,

title = "An LSTM-Based Method for Detection and Classification of Sensor Anomalies",

abstract = "Most existing machine learning (ML) based solutions for anomaly detection in sensory data rely on carefully hand-crafted features. This approach has a fundamental limitation since it is often application-specific and requires considerable human effort from domain experts. Deep learning models have been demonstrated to have the ability to abstract relevant high-level features from raw data. Long short-Term memory (LSTM) recurrent neural networks have proven effective in complex time-series prediction problems. In this paper, we propose an LSTM-based method for anomaly detection in sensory data. We systematically investigate its effectiveness on raw time-series of real medical sensors measurements and show that it achieves the same level of performance as traditional ML models operating on carefully designed feature vectors. The proposed method achieved micro, macro, and weighted precision, recall, and F1-score of over 0.99. ",

keywords = "Anomaly detection, deep learning, LSTM, machine learning, sensors validation, time-series analysis",

author = "Alexander Verner and Sumitra Mukherjee",

note = "Publisher Copyright: {\textcopyright} 2020 ACM.; 5th International Conference on Machine Learning Technologies, ICMLT 2020 ; Conference date: 19-06-2020 Through 21-06-2020",

year = "2020",

month = jun,

day = "19",

doi = "10.1145/3409073.3409089",

language = "English",

isbn = "9781450377645",

series = "Proceedings of the 2020 5th International Conference on Machine Learning Technologies",

publisher = "Association for Computing Machinery (ACM)",

pages = "39--45",

booktitle = "ICMLT 2020 - Proceedings of 2020 5th International Conference on Machine Learning Technologies",

}

TY - GEN

T1 - An LSTM-Based Method for Detection and Classification of Sensor Anomalies

AU - Verner, Alexander

AU - Mukherjee, Sumitra

PY - 2020/6/19

Y1 - 2020/6/19

N2 - Most existing machine learning (ML) based solutions for anomaly detection in sensory data rely on carefully hand-crafted features. This approach has a fundamental limitation since it is often application-specific and requires considerable human effort from domain experts. Deep learning models have been demonstrated to have the ability to abstract relevant high-level features from raw data. Long short-Term memory (LSTM) recurrent neural networks have proven effective in complex time-series prediction problems. In this paper, we propose an LSTM-based method for anomaly detection in sensory data. We systematically investigate its effectiveness on raw time-series of real medical sensors measurements and show that it achieves the same level of performance as traditional ML models operating on carefully designed feature vectors. The proposed method achieved micro, macro, and weighted precision, recall, and F1-score of over 0.99.

AB - Most existing machine learning (ML) based solutions for anomaly detection in sensory data rely on carefully hand-crafted features. This approach has a fundamental limitation since it is often application-specific and requires considerable human effort from domain experts. Deep learning models have been demonstrated to have the ability to abstract relevant high-level features from raw data. Long short-Term memory (LSTM) recurrent neural networks have proven effective in complex time-series prediction problems. In this paper, we propose an LSTM-based method for anomaly detection in sensory data. We systematically investigate its effectiveness on raw time-series of real medical sensors measurements and show that it achieves the same level of performance as traditional ML models operating on carefully designed feature vectors. The proposed method achieved micro, macro, and weighted precision, recall, and F1-score of over 0.99.

KW - Anomaly detection

KW - deep learning

KW - LSTM

KW - machine learning

KW - sensors validation

KW - time-series analysis

U2 - 10.1145/3409073.3409089

DO - 10.1145/3409073.3409089

M3 - Conference contribution

AN - SCOPUS:85090388640

SN - 9781450377645

T3 - Proceedings of the 2020 5th International Conference on Machine Learning Technologies

SP - 39

EP - 45

BT - ICMLT 2020 - Proceedings of 2020 5th International Conference on Machine Learning Technologies

PB - Association for Computing Machinery (ACM)

T2 - 5th International Conference on Machine Learning Technologies, ICMLT 2020

Y2 - 19 June 2020 through 21 June 2020

ER -

An LSTM-Based Method for Detection and Classification of Sensor Anomalies

Abstract

Publication series

Conference

Bibliographical note

ASJC Scopus Subject Areas

Keywords

Access to Document

Fingerprint

Cite this