In　an　effort　to　facilitate　alternative　splicing　discovery,　methods　for　facilitating　the　prediction　of　various　types　of　alternative　splicing　have　been　developed.　So　far,　efficient　computational　approach　has　not　been　developed　for　the　prediction　of　various　types　of　alternative　splicing　patterns　that　occur　widely　in　human　genome　as　well　as　in　other　organism.　Some　of　splicing　alternative　exons,　such　as　alternatively　spliced　internal　type,　have　some　combination　of　characteristics　which　make　it　easy　to　distinguish　from　constitutive　exons.　This　work　explores　the　use　of　EM　algorithm,　Gibbs　sampling　and　support　vector　machine　(SVM),　for　alternatively　spliced　internal　Exons　prediction.　Its　prediction　accuracy　was　evaluated　by　using　an　independent　set　of　exons　and　by　comparison　with　results　obtained　from　other　commonly　used　classification　method　(Logistic　regression　analysis)　using　the　same　dataset.　The　accuracy　of　the　prediction　is　83.6%,　which　is　comparable　to　the　results　obtained　by　other　classification　method.