چكيده لاتين :
Support Vector Machines (SVMs) and Multi-Resolution Analysis (MRA) both have been developed for solving signal approximation problem. Replacing the approximation criterion ofMRA by which be used in
SVMs, multi-resolution signal decomposition and approximation algorithm based on SVMs can be derived. The advantage of this algorithm not only reduces the approximation error by introducing structure risk, but also has better smoothness of approximation function. Experiment illustrates that this algorithm has better approximation performance than conventional MRA when applying it to the approximation of stationary signal.
