China
Africa
Explore chapters and articles related to this topic
Healthcare Applications Using Biomedical AI System
Published in Saravanan Krishnan, Ramesh Kesavan, B. Surendiran, G. S. Mahalakshmi, Handbook of Artificial Intelligence in Biomedical Engineering, 2021
S. Shyni Carmel Mary, S. Sasikala
Information technology and the biological system that has distributed communication nodes are stimulated neural networks. There are many algorithms used in a neural network. Very few are important, such as radial basis function network, perceptron, back propagation, logistic regression, gradient descent, and hop field network. Radial basis function network has an activation function called radial basis function in the hidden layer. It has an input layer, hidden layer, and linear output; also, it is used for the time series prediction, classification, and system control. Perceptron is a linear unsupervised binary classifier. There are two layers of perceptron such as a single layer and a multilayer. The multilayer is called the neural network. It has input layer weights and bias, net sum, and activation function.
Radial Basis Function Networks
Published in Sing-Tze Bow, Pattern Recognition and Image Preprocessing, 2002
As mentioned, the RBF method is developed from the exact interpolation approach, but with modifications, to provide a smooth interpolating function. The construction of a radial basis function network involves three different layers, namely, an input layer, a hidden layer, and an output layer. The input layer is primarily made up of source nodes (or sensory units) to hold the input data for processing. For an RBF in its basic form, there is only one hidden layer. This hidden layer is of high enough dimensions. It provides a nonlinear transformation from the input space. The output layer, which gives the network response to an activation pattern applied to the input layer, provides a linear transformation from the hidden unit space to the output space. Figure 9.1 shows the transformations imposed on the input vector by each layer of the RBF network. It can be noted that a nonlinear mapping is used to transform a nonlinearly separable classification problem into a linearly separable one.
Neural Networks and Adaptive Signal Processing
Published in Richard C. Dorf, Circuits, Signals, and Speech and Image Processing, 2018
Jose C. Principe, Mohamed Ibnkahla, Ahmad Iyanda Sulyman, Yu Cao
The radial basis function network is also a layered net with the hidden layer built from Gaussian kernels and a linear (or nonlinear) output layer (Figure 22.8). Training of the RBF network done normally in two stages (Principe et al., 2000): first, the centers xi are adaptively placed in the input space using competitive learning or k means clustering (Bishop, 1995), which are unsupervised procedures. Competitive learning is explained later in the chapter. The variances of each Gaussian is chosen as a percentage (30 to 50%) to the distance to the nearest center. The goal is to cover adequately the input data distribution. Once the RBF is located, the second layer weights wi are trained using the LMS procedure.
A traffic monitoring stream-based real-time vehicular offence detection approach
Published in Journal of Intelligent Transportation Systems, 2018
Automatic moving object detection, which segments moving objects from video streams, is a key technology in intelligent transportation systems (Huang & Chen, 2013; Oreifej, Li, & Shah, 2013). It is the preprocessing step for license plate detection from the traffic monitoring video stream (Cheng, Chen, & Huang, 2015). Moving object detection methods can be classified into three groups (Huang & Chen, 2014): temporal differencing, optical flow, and background subtraction. Background subtraction methods accomplish motion detection by comparing pixel features, which make the incoming image different from the reference background model of the previous images (Chen & Huang, 2015; Chen & Huang, 2014; Cheng, Chen, & Huang, 2015; Guo et al., 2013; Huang & Chen, 2014; Huang & Chen, 2013). Huang et al. (Chen & Huang, 2015; Chen & Huang, 2014; Huang & Chen, 2014; Huang & Chen, 2013) proposed methods for moving object detection from traffic monitoring video streams. The experimental results demonstrated that they can meet the requirement of real-time moving object detection. These methods solved the bandwidth limitation by a novel rate control scheme that alters the bit-rate to match the obtained network bandwidth. The article by Chen and Huang (2014) proposed to utilize a radial basis function network as its principle components.