Lakhdar Remaki

Lakhdar Remaki

Support vector machine (SVM), is a popular kernel method for data classification that demonstrated its efficiency for a large range of practical applications. The method suffers, however, from some weaknesses including; time processing, risk of failure of the optimization process for high dimension cases, generalization to multi-classes, unbalanced classes, and dynamic classification. In this paper an alternative method is proposed having a similar performance, with a sensitive improvement of the aforementioned shortcomings. The new method is based on a minimum distance to optimal subspaces containing the mapped original classes.

Lakhdar Remaki

This paper deals with the waves speed averaging impact impact on Godunov type schemes for linear scalar hyperbolic equations with discontinuous coefficients. In many numerical schemes of Godunov type used in fluid dynamics, electromagnetic, electro-hydrodynamic problems and so on, usually a Riemann problem needs to be solved to estimate fluxes. The exact solution is generally not possible to obtain, but good approximations are provided in many situations like Roe and HLLC Riemann solvers in fluids. However all these solvers assume that the acoustic waves speed are continuous by considering some averaging. This could unfortunately lead to a wrong solution as we will show in this paper for the linear scalar case. Providing a Riemann solver in the general case of non-linear hyperbolic systems with discontinuous waves speed is a very hard task, therefore in this paper and as a first step, we focus on the linear and scalar case. In a previous work we proposed for such problems a Riemann solution that takes into account the discontinuities of the waves speed, we provided a numerical argument to show the validity of the solution. In this paper, first a new argument using regularization technique is provided to reinforce the validity of the proposed solution. Then, the corresponding Godunov scheme is derived and the effect of waves speed averaging is clearly demonstrated with a clear connection to the distribution product phenomenon.