POSTER: Accuracy vs. Time Cost: Detecting Android Malware through Pareto Ensemble Pruning

摘要: This paper proposes Begonia, a malware detection system through Pareto ensemble pruning. We convert the problem into bi-objective optimization, aiming to trade off classification accuracy and size of classifiers as two objectives. automatically generate several groups base using SVM solutions optimization. then select ensembles with highest each group form final solutions, among which we hit optimal solution where combined loss function is minimal considering trade-off between time cost. expect users provide different levels their requirements best solution. Experimental results show that Begonia can achieve higher relatively lower overhead compared containing all make good requirements.