Supervised Deep Restricted Boltzmann Machine for Estimation of Concrete

Costly and time-consuming destructive methods are usually used to determine the properties of alternative concrete mixtures. To reduce cost and time, statistical and neural network models have been proposed to estimate concrete properties on the basis of input parameters. In this paper, a novel deep restricted Boltzmann machine is presented for estimating concrete properties based on mixture proportions. The effectiveness of the model is compared with two other widely used neural network/machine learning models: backpropagation neural network and support vector machine. The model is tested using 103 concrete test data from the machine learning repository of the University of California, Irvine. It is shown that the proposed model provides more accurate results and is computationally more efficient than the other two models. An accuracy of 98.0% is achieved for the example presented.