Reconfigurable Hardware Architecture of a Shape Recognition System Based on Specialized Tiny Neural Networks With Online Training

Abstract

Abstract-Neural networks are widely used in pattern recognition, security applications, and robot control. We propose a hardware architecture system using tiny neural networks (TNNs) specialized in image recognition. The generic TNN architecture allows for expandability by means of mapping several basic units (layers) and dynamic reconflguration, depending on the application speciflc demands. One of the most important features of TNNs is their learning ability. Weight modiflcation and architecture reconflguration can be carried out at run-time. Our system performs objects identiflcation by the interpretation of characteristics elements of their shapes. This is achieved by interconnecting several specialized TNNs. The results of several tests in different conditions are reported in this paper. The system accurately detects a test shape in most of the experiments performed. This paper also contains a detailed description of the system architecture and the processing steps. In order to valídate the research, the system has been implemented and conflgured as a perceptron network with back-propagation learning, choosing as reference application the recognition of shapes. Simulation results show that this architecture has signiflcant performance beneflts.

A reconfigurable architecture for real-time digital simulation of neurons

“…Moreno et al [10] present a digital approach in which they implement the perceptron model (a non-spiking, rate-code model), using an FPGA platform. They design VHDL models to implement tiny neural networks (TNN) for computer vision.…”

Research on the Models of Coupling Dynamics and Damage Classification for Vehicle-Engine Vibration

“…Besides the analysis of model coupling dynamics, this paper also studied the rotor damage detection. ere are several intelligent methods that have been used, such as the BP neural network method [18][19][20][21], the radical basis function (RBF) neural network method [22][23][24], genetic algorithm [25], the decision tree method [26], the k-nearest neighbour method [27], and the support vector machine [28]. Among the above methods, the BP method is a kind of multilayered feed-forward neural network, which includes the signal forward transfer and the error backward propagation.…”