Deception Detection using Random Forest-based Ensemble Learning

AbstractThe purpose of this work is to detect people lying using different ensemble machine learning algorithms to conclude a better classification model through comparison. Random Forest (RF) did an efficient work while dealing with both classification and regression problems; In this paper, we proposed a Random Forest-based ensemble learning, which is the combination of RF with SVM, GLM, KNNs, and GBM to improve the model performance. The data set that we used to fit into the machine learning models is Miami University Deception Detection Database (MU3D). MU3D is a free resource containing 320 videos of Black and White targets, female and male, telling truths and lies. We fit the MU3D video level data set into Random Forest-based ensemble learning models, which includes RF + SVM.Linear, RF + SVM.Poly, RF + GLM, RF + KNNs, RF + GBM (Stochastic Gradient Boosting) and RF + WSRF (Weighted Subspace Random Forest). As a comprehensive comparison of the model performance, we conclude our new combination of algorithms performs better than the traditional machine learning models. Our contribution in this work provides a robust classification method which improves the predicted performance while avoiding model overfitting.