Abstract:In data mining, identifying the best individual technique to achieve very reliable and accurate classification has always been considered as an important but non-trivial task. This paper presents a novel approachheterogeneous ensemble technique, to avoid the task and also to increase the accuracy of classification. It combines the models that are generated by using methodologically different learning algorithms and selected with different rules of utilizing both accuracy of individual modules and also diversity among the models. The key strategy is to select the most accurate model among all the generated models as the core model, and then select a number of models that are more diverse from the most accurate model to build the heterogeneous ensemble. The framework of the proposed approach has been implemented and tested on a real-world data to classify imaginary scenes. The results show our approach outperforms other the state of the art methods, including Bayesian network, SVM and AdaBoost.

When dealing with big data, "divide and conquer" is the most commonly used strategy in practice to partition a big dataset into such smaller subsets that each subset can be handled by a computer or a node of cluster or cloud computing systems. However, among many existing partitioning or sampling techniques, it is not clear which one is suitable and how the size of subset may affect the performance of further analysis. In this paper, after presenting a generic framework of ensemble approach for learning from big data, we focus our investigations on systematically evaluating the effect of partitioning strategies and subset size on ensemble performance. The experimental results have demonstrated that three investigated partitioning / sampling strategies behaved statistically similar but the subset size may affect the performance of the ensemble in very drastically different ways, which are grouped into three patterns, rather than just one default perception -the bigger the better.

No abstract

No abstract