Abstract
In this work, we propose a Spiking Neural Network (SNN) consisting of input neurons sparsely connected by plastic synapses to a randomly interlinked liquid, referred to as Liquid-SNN, for unsupervised speech and image recognition. We adapt the strength of the synapses interconnecting the input and liquid using Spike Timing Dependent Plasticity (STDP), which enables the neurons to self-learn a general representation of unique classes of input patterns. The presented unsupervised learning methodology makes it possible to infer the class of a test input directly using the liquid neuronal spiking activity. This is in contrast to standard Liquid State Machines (LSMs) that have fixed synaptic connections between the input and liquid followed by a readout layer (trained in a supervised manner) to extract the liquid states and infer the class of the input patterns. Moreover, the utility of LSMs has primarily been demonstrated for speech recognition. We find that training such LSMs is challenging for complex pattern recognition tasks because of the information loss incurred by using fixed input to liquid synaptic connections. We show that our Liquid-SNN is capable of efficiently recognizing both speech and image patterns by learning the rich temporal information contained in the respective input patterns. However, the need to enlarge the liquid for improving the accuracy introduces scalability challenges and training inefficiencies. We propose SpiLinC that is composed of an ensemble of multiple liquids operating in parallel. We use a “divide and learn” strategy for SpiLinC, where each liquid is trained on a unique segment of the input patterns that causes the neurons to self-learn distinctive input features. SpiLinC effectively recognizes a test pattern by combining the spiking activity of the constituent liquids, each of which identifies characteristic input features. As a result, SpiLinC offers competitive classification accuracy compared to the Liquid-SNN with added sparsity in synaptic connectivity and faster training convergence, both of which lead to improved energy efficiency in neuromorphic hardware implementations. We validate the efficacy of the proposed Liquid-SNN and SpiLinC on the entire digit subset of the TI46 speech corpus and handwritten digits from the MNIST dataset.
Highlights
Spiking Neural Network (SNN) are a class of bio-inspired neuromorphic computing paradigm that closely emulate the organization and computational efficiency of the human brain for complex classification and recognition tasks
We propose a general computing model referred to as the Liquid-SNN, consisting of input neurons sparsely connected by plastic synapses to a reservoir of spiking neurons termed liquid, for unsupervised learning of both speech and image patterns
We show that the presented Liquid-SNN, trained in an unsupervised manner on a subset of spoken digits from the TI46 speech corpus (Liberman et al, 1993), achieves comparable accuracy to that provided by Liquid State Machines (LSMs) trained using supervised algorithms
Summary
SNNs are a class of bio-inspired neuromorphic computing paradigm that closely emulate the organization and computational efficiency of the human brain for complex classification and recognition tasks. Several SNN architectures have been independently proposed for learning visual and auditory signal modalities. Two-layered fully-connected SNN (Diehl and Cook, 2015) and shallow/deep convolutional SNN (Masquelier and Thorpe, 2007; Lee et al, 2016, 2018a,b; Panda and Roy, 2016; Tavanaei et al, 2016; Panda et al, 2017a; Ferré et al, 2018; Jin et al, 2018; Kheradpisheh et al, 2018; Thiele et al, 2018; Wu et al, 2018) have been demonstrated for visual image recognition. It is highly desirable to have a general computing model capable of processing different signal modalities using a uniform self-learning methodology
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
