This　study　proposed　an　adaptive　mutation　particle　swarm　optimization　(AMPSO)　for　parameters　optimization　of　support　vector　machines　(SVM).　The　improved　inertia　weight　and　mutation　mechanism　aimed　to　balance　the　global　and　local　search,　which　could　improve　the　recognition　accuracy　of　SVM.　The　experimental　results　showed　that　compared　with　grid　and　particle　swarm　optimization　(PSO),　classification　accuracy　of　the　proposed　AMPSO-SVM　model　can　be　significantly　increased.