Recursive feature elimination

SVM-RFE Application of Optimal Brain Damage and Recursive Feature Elimination... 

Nowadays, as the rapid development of data collection techniques, more and more data are collected, and wrapper methods cannot meet the need of rapid data processing. Then, a number of input pruning methods using SVM have been developed. Guyon and his co-workers proposed to use optimal brain damage as the feature evaluation method and combined it with recursive feature elimination to perform gene selection [64],... 

Recursive feature elimination (RFE) is a heuristic method to search feature subset, which can remove the irrelevant features efficiently, and is adequate to be used in the data sets whose relevant features are relatively small. Guyon combined RFE with OBD to perform gene selection and proposed SVM-RFE algorithm. 

Selecting an optimum group of descriptors is both an important and time-consuming phase in developing a predictive QSAR model. Frohlich, Wegner, and Zell introduced the incremental regularized risk minimization procedure for SVM classification and regression models, and they compared it with recursive feature elimination and with the mutual information procedure. Their first experiment considered 164 compounds that had been tested for their human intestinal absorption, whereas the second experiment modeled the aqueous solubility prediction for 1297 compounds. Structural descriptors were computed by those authors with JOELib and MOE, and full cross-validation was performed to compare the descriptor selection methods. The incremental... 

Equations (5-42) to (5-44) constitute a set of simultaneous linear equations in the unknown values of the quantity on the new profile, each of the sets for temperature and the conversions being independent. The matrix of coefficients is the same for each conversion, and differs for the temperature only in one element, the one containing the heat-transfer coefficient. An important property of each matrix of coefficients is that it is independent of the axial position, so that for the purpose of the calculation it is a constant matrix. Another important property is that only three diagonals of the matrix contain elements that are not zero. These properties make it possible to throw the calculation of the unknown quantities into a very simple form. The essential feature of the calculation is that coefficients in two two-term recursion formulas are constructed from the matrix elements, and then these coefficients are used to calculate, first, a set of ancillary quantities, and then the desired quantities. The procedure is exactly what would be used in eliminating the unknown quantities successively from the equations and then sub-... 

---