Abstract
Precise spike timing is considered to play a fundamental role in communications and signal processing in biological neural networks. Understanding the mechanism of spike timing adjustment would deepen our understanding of biological systems and enable advanced engineering applications such as efficient computational architectures. However, the biological mechanisms that adjust and maintain spike timing remain unclear. Existing algorithms adopt a supervised approach, which adjusts the axonal conduction delay and synaptic efficacy until the spike timings approximate the desired timings. This study proposes a spike timing-dependent learning model that adjusts the axonal conduction delay and synaptic efficacy in both unsupervised and supervised manners. The proposed learning algorithm approximates the Expectation-Maximization algorithm, and classifies the input data encoded into spatio-temporal spike patterns. Even in the supervised classification, the algorithm requires no external spikes indicating the desired spike timings unlike existing algorithms. Furthermore, because the algorithm is consistent with biological models and hypotheses found in existing biological studies, it could capture the mechanism underlying biological delay learning.
Highlights
As confirmed in biological studies, the precise timing of a neuronal spike plays a fundamental role in information processing in the central nervous system (Carr and Konishi, 1990; Middlebrooks et al, 1994; Seidl et al, 2010)
The spiking neural network (SNN) and the proposed learning algorithm based on the Multinoulli-Bernoulli mixture model were evaluated on toy spike patterns generated by three pre-synaptic neurons
The proposed learning algorithm is adaptable to supervised learning, in which synaptic plasticity and delay learning are not self-driven by generated post-synaptic spikes, but are driven by external spikes zt = 1 generated at the desired timings (Bohte et al, 2002; Ponulak, 2005; Gütig and Sompolinsky, 2006; Pfister et al, 2006; Paugam-Moisy et al, 2008; Taherkhani et al, 2015; Matsubara and Torikai, 2016)
Summary
As confirmed in biological studies, the precise timing of a neuronal spike plays a fundamental role in information processing in the central nervous system (Carr and Konishi, 1990; Middlebrooks et al, 1994; Seidl et al, 2010). Information processing in a neural circuit requires the synchronous arrival of spikes elicited by multiple pre-synaptic neurons, and optimal conduction delay in the axons is critical. Supervised delay learning algorithms such as the DLReSuMe algorithm (Taherkhani et al, 2015) directly adjust the conduction delay to suit the given spatio-temporal patterns of the pre- and post-synaptic spikes. These algorithms answer the purpose to reproduce the spatio-temporal patterns, but are not suited for classification of the spatio-temporal patterns because the desired timings are generally unknown and should be manually adjusted with great care in the classification tasks This approach reduces the flexibility in the context of machine learning and poorly represents biological systems that selfadapt to changing environments. The present study proposes an unsupervised learning algorithm that adjusts the conduction delays and synaptic weights of an SNN. Preliminary and limited results of this study were presented in our conference paper (Matsubara, 2017)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
