A robust approach through combining optimized neural network and optimized support vector regression for modeling deformation modulus of rock masses

Deformation modulus is regarded as one of the fundamental parameters in design and well performing of rock engineering tasks. Measurement of this parameter in laboratory usually require a lot of effort, time and money. Accordingly, search to find fast, robust, and dependable model for estimation of this parameter is obligatory. In first stage of this communication, two improved mathematical approaches including optimized neural network (ONN), and optimized support vector regression (OSVR) are applied to estimate deformation modulus of rock masses. Outputs of optimized models are then integrated by virtue of committee machine (CM). Combination of optimized model erected model which achieve advantage of individual models. Optimization model which used for optimizing the predictive models as well as determining the optimal contribution of optimized model in CM is particle swarm optimization. In order to reveal the supremacy of the CM model over its rivals (ONN, and OSVR) error analysis based on statistical criteria is applied. Results of this study deduced that CM model has superior performance in predicting the deformation modulus of rock masses.