Abstract
Spatiotemporal patterns, such as words in speech, are rarely precisely the same duration, yet a word spoken faster or slower is still easily recognisable. Neural mechanisms underlying this ability to recognise stretched or compressed versions of the same spatiotemporal pattern are not well understood. Recognition of time-varying patterns is often studied at the network level, however here we propose a single neuron using learnable spike delays for the task. We characterise the response of a single neuron to stretched and compressed versions of a learnt pattern and show that using delays leads to pattern recognition above 99% accuracy for patterns morphed by up to 50%. Additionally, we demonstrate a significantly reduced response when the pattern is reversed, a property that is often difficult to reproduce in synaptic efficacy (synaptic weight) based learning systems. With appropriate settings of the neuron membrane time constant and spike threshold, we show that a single neuron is able to generalise to time-warped patterns while discriminating temporally reversed patterns. Together, these results highlight the potential of synaptic delay-based learning rules as a robust mechanism for learning time-warped spatiotemporal patterns.
This project was supported by a UQ scholarship to J.A. and funding from the ARC Centre of Excellence for the Dynamics of Language Project ID: CE140100041.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Afshar, S., George, L., Tapson, J., van Schaik, A., Hamilton, T.J.: Racing to learn: statistical inference and learning in a single spiking neuron with adaptive kernels. Front. Neuroscience 8 (2014). https://doi.org/10.3389/fnins.2014.00377
Arnold, J., Statton, P., Wiles, J.: Conduction delay plasticity can robustly learn spatiotemporal patterns embedded in noise. In: The 2021 International Joint Conference on Neural Networks (IJCNN) (in press)
Caporale, N., Dan, Y.: Spike timing-dependent plasticity: a Hebbian learning rule. Annu. Rev. Neurosci. 31(1), 25–46 (2008). https://doi.org/10.1146/annurev.neuro.31.060407.125639
Carr, C.E., Konishi, M.: Axonal delay lines for time measurement in the owl’s brainstem. Proc. Natl. Acad. Sci. 85(21), 8311–8315 (1988). https://doi.org/10.1073/pnas.85.21.8311
Fields, R.D.: Myelination: an overlooked mechanism of synaptic plasticity? Neuroscientist 11(6), 528–531 (2005). https://doi.org/10.1177/1073858405282304
Fields, R.D.: A new mechanism of nervous system plasticity: activity-dependent myelination. Nat. Rev. Neurosci. 16(12), 756–767 (2015). https://doi.org/10.1038/nrn4023
Gütig, R., Sompolinsky, H.: The tempotron: a neuron that learns spike timing-based decisions. Nat. Neurosci. 9(3), 420–428 (2006). https://doi.org/10.1038/nn1643
König, P., Engel, A.K., Singer, W.: Integrator or coincidence detector? The role of the cortical neuron revisited. Trends Neurosci. 19(4), 130–137 (1996). https://doi.org/10.1016/S0166-2236(96)80019-1
Masquelier, T.: STDP allows close-to-optimal spatiotemporal spike pattern detection by single coincidence detector neurons. Neuroscience 389, 133–140 (2018). https://doi.org/10.1016/j.neuroscience.2017.06.032
Masquelier, T., Thorpe, S.J.: Unsupervised learning of visual features through spike timing dependent plasticity. PLOS Comput. Biol. 3(2), e31 (2007). https://doi.org/10.1371/journal.pcbi.0030031
Matsubara, T.: Conduction delay learning model for unsupervised and supervised classification of spatio-temporal spike patterns. Front. Comput. Neurosci 11 (2017). https://doi.org/10.3389/fncom.2017.00104
Natschläger, T., Ruf, B.: Spatial and temporal pattern analysis via spiking neurons. Netw. Comput. Neural Syst. 9(3), 319–332 (1998). https://doi.org/10.1088/0954-898X/9/3/003
Tank, D.W., Hopfield, J.J.: Neural computation by concentrating information in time. Proc. Natl. Acad. Sci. U.S.A. 84(7), 1896–1900 (1987)
Wright, P.W., Wiles, J.: Learning transmission delays in spiking neural networks: a novel approach to sequence learning based on spike delay variance. In: The 2012 International Joint Conference on Neural Networks (IJCNN), pp. 1–8 (2012). https://doi.org/10.1109/IJCNN.2012.6252371
Wright, P.: Exploring the impact of interneurons and delay learning on temporal spike computation. B.E. Thesis, The University of Queensland, Brisbane, Australia (2011)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Arnold, J., Stratton, P., Wiles, J. (2021). Single Neurons with Delay-Based Learning Can Generalise Between Time-Warped Patterns. In: Farkaš, I., Masulli, P., Otte, S., Wermter, S. (eds) Artificial Neural Networks and Machine Learning – ICANN 2021. ICANN 2021. Lecture Notes in Computer Science(), vol 12894. Springer, Cham. https://doi.org/10.1007/978-3-030-86380-7_11
Download citation
DOI: https://doi.org/10.1007/978-3-030-86380-7_11
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-86379-1
Online ISBN: 978-3-030-86380-7
eBook Packages: Computer ScienceComputer Science (R0)