Support Vector Machines (SVM) represent one of the leading methods used to address pattern recognition and machine learning problems.

They have found application in several systems biology domains too, despite their high dimensionality and complex dynamics.

One goal is to optimize their performance by:

• improving robustness and efficiency (through a calibrated input selection process according to relevance, information content etc.);
• enabling sparsity (reduction of kernel functions, weight pruning, shrinkage etc.)
• generating  greediness (like in many sequential algorithms)

Relevance Vector Machines represent an example of an effective method that refines SVM.