Autonomous Learning Paradigm for Spiking Neural Networks

Junxiu Liu, Liam McDaid, Jim Harkin, Shvan Haji Karim, Anju P. Johnson, David M. Halliday, Andy Tyrrell, Jon Timmis, Alan G. Millard, James Hilder

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Compared to biological systems, existing learning systems lack the ability to learn autonomously, especially in changing and dynamic environments. This paper addresses the issue of autonomous learning by developing a self-learning spiking neural network (SNN) and demonstrating its autonomous learning capability using a simple robot controller application. Our proposed learning rule exploits an inherit property of the existing Spike-Timing-Dependent Plasticity (STDP) rule in that if the instantaneous presynaptic frequency decreases, then for a conventional Hebbian window the STDP rule potentiates. Conversely if the instantaneous frequency increases the STDP rule depresses: the opposite is true for anti-Hebbian window. This paper will also show that obstacle avoidance is achievable using a conventional Hebbian learning window while object tracking can be learned using an anti-Hebbian learning window. Hence the proposed learning paradigm is novel in that it does not require external supervisions for either these tasks. The proposed learning paradigm also uses a previously explored astrocyte neuron interaction where a periodic Slow Inward Current (SIC) from an astrocyte can potentiate a postsynaptic neuron for a period of time: this time window can be used to strengthen/weaken synaptic pathways. An obstacle avoidance task is used for the performance analysis and results show that the SNN based robot controller has autonomous learning capabilities under the dynamic conditions.

LanguageEnglish
Title of host publicationArtificial Neural Networks and Machine Learning – ICANN 2019
Subtitle of host publicationTheoretical Neural Computation - 28th International Conference on Artificial Neural Networks, 2019, Proceedings
EditorsIgor V. Tetko, Pavel Karpov, Fabian Theis, Vera Kurková
Pages737-744
Number of pages8
Volume11727
ISBN (Electronic)978-3-030-30487-4
DOIs
Publication statusPublished - 9 Sep 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11727 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Fingerprint

Spiking Neural Networks
Plasticity
Paradigm
Collision avoidance
Neural networks
Neurons
Spike
Hebbian Learning
Instantaneous Frequency
Timing
Robots
Obstacle Avoidance
Controllers
Biological systems
Dependent
Neuron
Robot
Learning systems
Controller
Self-learning

Keywords

  • Learning
  • Plasticity windows
  • Robots
  • SNN

Cite this

Liu, J., McDaid, L., Harkin, J., Haji Karim, S., Johnson, A. P., Halliday, D. M., ... Hilder, J. (2019). Autonomous Learning Paradigm for Spiking Neural Networks. In I. V. Tetko, P. Karpov, F. Theis, & V. Kurková (Eds.), Artificial Neural Networks and Machine Learning – ICANN 2019: Theoretical Neural Computation - 28th International Conference on Artificial Neural Networks, 2019, Proceedings (Vol. 11727, pp. 737-744). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11727 LNCS). https://doi.org/10.1007/978-3-030-30487-4_57
Liu, Junxiu ; McDaid, Liam ; Harkin, Jim ; Haji Karim, Shvan ; Johnson, Anju P. ; Halliday, David M. ; Tyrrell, Andy ; Timmis, Jon ; Millard, Alan G. ; Hilder, James. / Autonomous Learning Paradigm for Spiking Neural Networks. Artificial Neural Networks and Machine Learning – ICANN 2019: Theoretical Neural Computation - 28th International Conference on Artificial Neural Networks, 2019, Proceedings. editor / Igor V. Tetko ; Pavel Karpov ; Fabian Theis ; Vera Kurková. Vol. 11727 2019. pp. 737-744 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{abd1cda0eae344c2afe6a47e20871ad8,
title = "Autonomous Learning Paradigm for Spiking Neural Networks",
abstract = "Compared to biological systems, existing learning systems lack the ability to learn autonomously, especially in changing and dynamic environments. This paper addresses the issue of autonomous learning by developing a self-learning spiking neural network (SNN) and demonstrating its autonomous learning capability using a simple robot controller application. Our proposed learning rule exploits an inherit property of the existing Spike-Timing-Dependent Plasticity (STDP) rule in that if the instantaneous presynaptic frequency decreases, then for a conventional Hebbian window the STDP rule potentiates. Conversely if the instantaneous frequency increases the STDP rule depresses: the opposite is true for anti-Hebbian window. This paper will also show that obstacle avoidance is achievable using a conventional Hebbian learning window while object tracking can be learned using an anti-Hebbian learning window. Hence the proposed learning paradigm is novel in that it does not require external supervisions for either these tasks. The proposed learning paradigm also uses a previously explored astrocyte neuron interaction where a periodic Slow Inward Current (SIC) from an astrocyte can potentiate a postsynaptic neuron for a period of time: this time window can be used to strengthen/weaken synaptic pathways. An obstacle avoidance task is used for the performance analysis and results show that the SNN based robot controller has autonomous learning capabilities under the dynamic conditions.",
keywords = "Learning, Plasticity windows, Robots, SNN",
author = "Junxiu Liu and Liam McDaid and Jim Harkin and {Haji Karim}, Shvan and Johnson, {Anju P.} and Halliday, {David M.} and Andy Tyrrell and Jon Timmis and Millard, {Alan G.} and James Hilder",
year = "2019",
month = "9",
day = "9",
doi = "10.1007/978-3-030-30487-4_57",
language = "English",
isbn = "978-3-030-30486-7",
volume = "11727",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "737--744",
editor = "Tetko, {Igor V.} and Pavel Karpov and Fabian Theis and Vera Kurkov{\'a}",
booktitle = "Artificial Neural Networks and Machine Learning – ICANN 2019",

}

Liu, J, McDaid, L, Harkin, J, Haji Karim, S, Johnson, AP, Halliday, DM, Tyrrell, A, Timmis, J, Millard, AG & Hilder, J 2019, Autonomous Learning Paradigm for Spiking Neural Networks. in IV Tetko, P Karpov, F Theis & V Kurková (eds), Artificial Neural Networks and Machine Learning – ICANN 2019: Theoretical Neural Computation - 28th International Conference on Artificial Neural Networks, 2019, Proceedings. vol. 11727, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11727 LNCS, pp. 737-744. https://doi.org/10.1007/978-3-030-30487-4_57

Autonomous Learning Paradigm for Spiking Neural Networks. / Liu, Junxiu; McDaid, Liam; Harkin, Jim; Haji Karim, Shvan; Johnson, Anju P.; Halliday, David M.; Tyrrell, Andy; Timmis, Jon; Millard, Alan G.; Hilder, James.

Artificial Neural Networks and Machine Learning – ICANN 2019: Theoretical Neural Computation - 28th International Conference on Artificial Neural Networks, 2019, Proceedings. ed. / Igor V. Tetko; Pavel Karpov; Fabian Theis; Vera Kurková. Vol. 11727 2019. p. 737-744 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11727 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Autonomous Learning Paradigm for Spiking Neural Networks

AU - Liu, Junxiu

AU - McDaid, Liam

AU - Harkin, Jim

AU - Haji Karim, Shvan

AU - Johnson, Anju P.

AU - Halliday, David M.

AU - Tyrrell, Andy

AU - Timmis, Jon

AU - Millard, Alan G.

AU - Hilder, James

PY - 2019/9/9

Y1 - 2019/9/9

N2 - Compared to biological systems, existing learning systems lack the ability to learn autonomously, especially in changing and dynamic environments. This paper addresses the issue of autonomous learning by developing a self-learning spiking neural network (SNN) and demonstrating its autonomous learning capability using a simple robot controller application. Our proposed learning rule exploits an inherit property of the existing Spike-Timing-Dependent Plasticity (STDP) rule in that if the instantaneous presynaptic frequency decreases, then for a conventional Hebbian window the STDP rule potentiates. Conversely if the instantaneous frequency increases the STDP rule depresses: the opposite is true for anti-Hebbian window. This paper will also show that obstacle avoidance is achievable using a conventional Hebbian learning window while object tracking can be learned using an anti-Hebbian learning window. Hence the proposed learning paradigm is novel in that it does not require external supervisions for either these tasks. The proposed learning paradigm also uses a previously explored astrocyte neuron interaction where a periodic Slow Inward Current (SIC) from an astrocyte can potentiate a postsynaptic neuron for a period of time: this time window can be used to strengthen/weaken synaptic pathways. An obstacle avoidance task is used for the performance analysis and results show that the SNN based robot controller has autonomous learning capabilities under the dynamic conditions.

AB - Compared to biological systems, existing learning systems lack the ability to learn autonomously, especially in changing and dynamic environments. This paper addresses the issue of autonomous learning by developing a self-learning spiking neural network (SNN) and demonstrating its autonomous learning capability using a simple robot controller application. Our proposed learning rule exploits an inherit property of the existing Spike-Timing-Dependent Plasticity (STDP) rule in that if the instantaneous presynaptic frequency decreases, then for a conventional Hebbian window the STDP rule potentiates. Conversely if the instantaneous frequency increases the STDP rule depresses: the opposite is true for anti-Hebbian window. This paper will also show that obstacle avoidance is achievable using a conventional Hebbian learning window while object tracking can be learned using an anti-Hebbian learning window. Hence the proposed learning paradigm is novel in that it does not require external supervisions for either these tasks. The proposed learning paradigm also uses a previously explored astrocyte neuron interaction where a periodic Slow Inward Current (SIC) from an astrocyte can potentiate a postsynaptic neuron for a period of time: this time window can be used to strengthen/weaken synaptic pathways. An obstacle avoidance task is used for the performance analysis and results show that the SNN based robot controller has autonomous learning capabilities under the dynamic conditions.

KW - Learning

KW - Plasticity windows

KW - Robots

KW - SNN

UR - http://www.scopus.com/inward/record.url?scp=85072850927&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-30487-4_57

DO - 10.1007/978-3-030-30487-4_57

M3 - Conference contribution

SN - 978-3-030-30486-7

VL - 11727

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

SP - 737

EP - 744

BT - Artificial Neural Networks and Machine Learning – ICANN 2019

A2 - Tetko, Igor V.

A2 - Karpov, Pavel

A2 - Theis, Fabian

A2 - Kurková, Vera

ER -

Liu J, McDaid L, Harkin J, Haji Karim S, Johnson AP, Halliday DM et al. Autonomous Learning Paradigm for Spiking Neural Networks. In Tetko IV, Karpov P, Theis F, Kurková V, editors, Artificial Neural Networks and Machine Learning – ICANN 2019: Theoretical Neural Computation - 28th International Conference on Artificial Neural Networks, 2019, Proceedings. Vol. 11727. 2019. p. 737-744. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-30487-4_57