Chapter 3
Concrete Quality Prediction Using Deep Neural Networks
Section 2
Basics Concepts of Artificial Neural Networks (ANN’s)
ANNs are mathematical models that are able to simulate the usual activities of the human brain such as image perception, pattern recognition, language comprehension, and sensory motor coordination. These models are composed of a system of nodes, equivalent to the neurons of a human brain, which are interconnected by weighted connections, equivalent to the synapses between the neurons. The output of the network is iteratively changed from the link weights up to the convergence. The data to be analyzed is provided via the input level and the result provided by the network is returned from the output level. Input nodes represent the independent or predictive variables used to predict dependent variables, such as output neurons. Here we will learn about: - Basic concepts of ANN’s - Architecture of ANN’s - Paradigms - Supervised Learning - Unsupervised Learning - Semi-Supervised Learning - Understanding the structure of Neural networks - Weights and Biases - Types of activation functions - Unit step activation function - Sigmoid - Hyperbolic Tangent - Rectified Linear Unit (ReLU)