The ripple pond: enabling spiking networks to see

Abstract

We present the biologically inspired Ripple Pond Network (RPN), a simply connected spiking neural network which performs a transformation converting two dimensional images to one dimensional temporal patterns (TP) suitable for recognition by temporal coding learning and memory networks. The RPN has been developed as a hardware solution linking previously implemented neuromorphic vision and memory structures such as frameless vision sensors and neuromorphic temporal coding spiking neural networks. Working together such systems are potentially capable of delivering end-to-end high-speed, low-power and low-resolution recognition for mobile and autonomous applications where slow, highly sophisticated and power hungry signal processing solutions are ineffective. Key aspects in the proposed approach include utilizing the spatial properties of physically embedded neural networks and propagating waves of activity therein for information processing, using dimensional collapse of imagery information into amenable TP and the use of asynchronous frames for information binding.

Racing to learn: statistical inference and learning in a single spiking neuron with adaptive kernels

“…As a single neuron, SKAN has previously been shown to select and learn the most common spatio-pattern presented in a random sequence containing multiple patterns (Sofatzis et al, 2014a ). This effect has been demonstrated in the context of visual processing where hand gestures were transformed to spatio-temporal patterns via a neuronal transform operation (Afshar et al, 2013 ) and processed by SKAN (Sofatzis et al, 2014b ). Figure 7 shows the performance of a four input neuron as a function of spatio-temporal pattern probability.…”

Investigation of Event-Based Surfaces for High-Speed Detection, Unsupervised Feature Extraction, and Object Recognition

“…Figures 2C,D show the distribution in the number of events per recording and recording duration for the dataset. The full dataset can be found at Afshar et al (2018).…”

Rotationally invariant vision recognition with neuromorphic transformation and learning networks

“…As proposed by Afshar et al [1], a Ripple Pond Network (RPN) can be used as a critical 2D to 1D image conversion stage in a larger neural system that is capable of basic, yet effective, view-invariant recognition. Using the simplest implementation as per the work of this paper, such a system would consist of the following stages: image capturing, a static disc RPN, and a temporal coding memory network.…”

“…Stemming from recent research by Afshar et al [1] into the minimal neural networks of such simple animals, this paper presents an end-to-end neuromorphic vision recognition system implementing: the Ripple Pond Network (RPN), a neural network that transforms a 2D image to a 1D rotationally invariant temporal pattern (TPs); the Synaptic Delay Adaptation Network (SKAN), a neural network capable of unsupervised learning of a spatio-temporal pattern of input spikes [5]; and a bridging network that converts the TPs into a spiking pattern that can be learnt by the SKAN. The simplicity of this implementation is highlighted by the fact that it is built entirely on a low cost, low power, Field Programmable Gate Array (FPGA) and is capable of rapid learning and recognition of simple hand gestures (albeit with limited accuracy) with no prior training.…”