Two-Layered Oscillatory Neural Networks with Analog Feedforward Majority Gate for Image Edge Detection Application

Abstract

The increasing volume of smart edge devices, like smart cameras, and the growing amount of data to treat incited the development of light edge Artificial Intelligence (AI) solutions with neuromorphic computing. Oscillatory Neural Network (ONN) is a promising neuromorphic computing approach which uses networks of coupled oscillators, and their inherent parallel synchronization to compute. Also, ONN phase computing allows to limit voltage amplitude and reduce power consumption. Low-power, fast, and parallel computation properties make ONN attractive for edge AI. In state-of-theart, ONN is built with a fully-connected architecture, with coupling defined from unsupervised learning to perform autoassociative memory tasks, like with Hopfield Networks. However, to allow ONN to solve beyond associative memory applications, there is a need to explore further ONN architectures. In this work, we propose a novel architecture of cascaded analog fully-connected ONNs interconnected with an analog feedforward majority gate layer. In particular, we show this architecture can solve image edge detection task using two fully-connected ONN layers. This is, to our best knowledge, a first analog-based solution to cascade two fully-connected ONNs.

Multiple Sclerosis Recognition via Wavelet Entropy and PSO-Based Neural Network

“…Each neuron in a layer is connected to every neuron in the subsequent layer, but there are no connections that loop back within the same layer or to previous layers. Each connection between neurons has an associated weight, which is adjusted during training to enable the network to learn and make predictions [38]. The operation of a FNN involves a simple forward propagation of information.…”