Unsupervised Handwritten Signature Verification with Extreme Learning Machines

Anton Akusok; Leonardo Espinosa-Leal; Amaury Lendasse; Kaj-Mikael Björk

doi:10.1007/978-3-031-21678-7_12

Unsupervised Handwritten Signature Verification with Extreme Learning Machines

Anton Akusok^*, Leonardo Espinosa-Leal, Amaury Lendasse, Kaj-Mikael Björk

^*Corresponding author for this work

Not published at VTT

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

Abstract

Handwritten signature verification has two approaches based on online or offline data collection, both of them being supervised machine learning tasks. This work investigates the feasibility of unsupervised signature verification. It is inspired by a model-based forged signature generation approach, whose inversion could potentially provide an unsupervised solution for the signature verification task. The model inversion is attempted on a massive collection of image patches taken from samples of a large GPDSS10000 artificial signature verification dataset, pre-processed by a general-purpose deep learning network that extracts 1024 meaningful image features. An Extreme Learning Machine (ELM) solves the inversion problem at a very large scale. The paper proposes practical ways of ELM model structure selection on massive datasets and faster solvers. The results show the feasibility of an unsupervised solution for signature verification.

Original language	English
Title of host publication	Proceedings of ELM 2021
Subtitle of host publication	Theory, Algorithms and Applications
Editors	Kaj-Mikael Björk
Publisher	Springer
ISBN (Electronic)	978-3-031-21678-7
ISBN (Print)	978-3-031-21677-0, 978-3-031-21680-0
DOIs	https://doi.org/10.1007/978-3-031-21678-7_12
Publication status	Published - 2023
MoE publication type	A4 Article in a conference publication
Event	11th International Conference on Extreme Learning Machines (ELM2021) - On-line, Helsinki, Finland Duration: 15 Dec 2021 → 16 Dec 2021 Conference number: 11 https://risklab.fi/events/

Publication series

Series	Proceedings in Adaptation, Learning and Optimization
Volume	16
ISSN	2363-6084

Conference

Conference	11th International Conference on Extreme Learning Machines (ELM2021)
Abbreviated title	ELM2021
Country/Territory	Finland
City	Helsinki
Period	15/12/21 → 16/12/21
Internet address	https://risklab.fi/events/

Access to Document

10.1007/978-3-031-21678-7_12

Cite this

@inproceedings{f105afc145d74d2ea2f719b9dfdbbf6c,

title = "Unsupervised Handwritten Signature Verification with Extreme Learning Machines",

abstract = "Handwritten signature verification has two approaches based on online or offline data collection, both of them being supervised machine learning tasks. This work investigates the feasibility of unsupervised signature verification. It is inspired by a model-based forged signature generation approach, whose inversion could potentially provide an unsupervised solution for the signature verification task. The model inversion is attempted on a massive collection of image patches taken from samples of a large GPDSS10000 artificial signature verification dataset, pre-processed by a general-purpose deep learning network that extracts 1024 meaningful image features. An Extreme Learning Machine (ELM) solves the inversion problem at a very large scale. The paper proposes practical ways of ELM model structure selection on massive datasets and faster solvers. The results show the feasibility of an unsupervised solution for signature verification.",

author = "Anton Akusok and Leonardo Espinosa-Leal and Amaury Lendasse and Kaj-Mikael Bj{\"o}rk",

year = "2023",

doi = "10.1007/978-3-031-21678-7_12",

language = "English",

isbn = "978-3-031-21677-0",

series = "Proceedings in Adaptation, Learning and Optimization",

publisher = "Springer",

editor = "Kaj-Mikael Bj{\"o}rk",

booktitle = "Proceedings of ELM 2021",

address = "Germany",

note = "11th International Conference on Extreme Learning Machines (ELM2021), ELM2021 ; Conference date: 15-12-2021 Through 16-12-2021",

url = "https://risklab.fi/events/",

}

Akusok, A, Espinosa-Leal, L, Lendasse, A & Björk, K-M 2023, Unsupervised Handwritten Signature Verification with Extreme Learning Machines. in K-M Björk (ed.), Proceedings of ELM 2021: Theory, Algorithms and Applications. Springer, Proceedings in Adaptation, Learning and Optimization, vol. 16, 11th International Conference on Extreme Learning Machines (ELM2021), Helsinki, Finland, 15/12/21. https://doi.org/10.1007/978-3-031-21678-7_12

Unsupervised Handwritten Signature Verification with Extreme Learning Machines. / Akusok, Anton; Espinosa-Leal, Leonardo; Lendasse, Amaury et al.
Proceedings of ELM 2021: Theory, Algorithms and Applications. ed. / Kaj-Mikael Björk. Springer, 2023. (Proceedings in Adaptation, Learning and Optimization, Vol. 16).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Unsupervised Handwritten Signature Verification with Extreme Learning Machines

AU - Akusok, Anton

AU - Espinosa-Leal, Leonardo

AU - Lendasse, Amaury

AU - Björk, Kaj-Mikael

N1 - Conference code: 11

PY - 2023

Y1 - 2023

N2 - Handwritten signature verification has two approaches based on online or offline data collection, both of them being supervised machine learning tasks. This work investigates the feasibility of unsupervised signature verification. It is inspired by a model-based forged signature generation approach, whose inversion could potentially provide an unsupervised solution for the signature verification task. The model inversion is attempted on a massive collection of image patches taken from samples of a large GPDSS10000 artificial signature verification dataset, pre-processed by a general-purpose deep learning network that extracts 1024 meaningful image features. An Extreme Learning Machine (ELM) solves the inversion problem at a very large scale. The paper proposes practical ways of ELM model structure selection on massive datasets and faster solvers. The results show the feasibility of an unsupervised solution for signature verification.

AB - Handwritten signature verification has two approaches based on online or offline data collection, both of them being supervised machine learning tasks. This work investigates the feasibility of unsupervised signature verification. It is inspired by a model-based forged signature generation approach, whose inversion could potentially provide an unsupervised solution for the signature verification task. The model inversion is attempted on a massive collection of image patches taken from samples of a large GPDSS10000 artificial signature verification dataset, pre-processed by a general-purpose deep learning network that extracts 1024 meaningful image features. An Extreme Learning Machine (ELM) solves the inversion problem at a very large scale. The paper proposes practical ways of ELM model structure selection on massive datasets and faster solvers. The results show the feasibility of an unsupervised solution for signature verification.

U2 - 10.1007/978-3-031-21678-7_12

DO - 10.1007/978-3-031-21678-7_12

M3 - Conference article in proceedings

SN - 978-3-031-21677-0

SN - 978-3-031-21680-0

T3 - Proceedings in Adaptation, Learning and Optimization

BT - Proceedings of ELM 2021

A2 - Björk, Kaj-Mikael

PB - Springer

T2 - 11th International Conference on Extreme Learning Machines (ELM2021)

Y2 - 15 December 2021 through 16 December 2021