Inorganic Phosphates Investigation by Support Vector Machine

We dealt the prediction of crystal chemical features of new sinthesized inorganic phosphates with a supervised-learning regression problem. Then, we analysed correlations between crystal chemical properties of phosphate crystals by a learning machine method, Support Vector Machine (SVM), to develop the detection algorithm. Using structural properties of phosphate crystal structures widely described in the literature, we developed several SVMs able to capture statistical relations between crystal chemical properties of the anhydrous phosphates from the available dataset. In this way, we demonstrated the suitability of SVM for the prediction of structural properties of crystals.