I think you have now understood that one of the relevant points when applying any kind of data mining model is to know and test its assumptions, in order to understand how significant and reliable the conclusions you can draw from it are.

When talking about support vector machines, the only assumption usually cited to be relevant is the presence of independent and identically distributed random variables.