Browsing by Author "Jiang, Zhonghua"

Now showing 1 - 4 of 4
  • Thumbnail Image
    Item
    A Novel Regression Model Combining Instance Based Rule Mining With EM Algorithm
    (2013-04-01) Jiang, Zhonghua; Karypis, George
    In recent years, there have been increasing efforts to apply association rule mining to build Associative Classification (AC) models. However, the similar area that applies association rule mining to build Associative Regression (AR) models has not been well explored. In this work, we fill this gap by presenting a novel regression model based on association rules called AREM. AREM derives a set of regression rules by: (i) applying an instance based approach to mine itemsets which form the regression rules' left hand side, and (ii) developing a probabilistic model which determines, for each mined itemset, the corresponding rule's right hand side and the importance weight. To address the computational bottleneck of the traditional two-step approach for itemset mining, AREM utilizes an Instance-Based Itemset Miner (IBIMiner) algorithm that directly discovers the final set of itemsets. IBIMiner incorporates various methods to bound the quality of any future extensions of the itemset under consideration. These bounds are then used to prune the search space. In addition, AREM treats the regression rules' right hand side and importance weights as parameters of a probabilistic model, which are then learned in the expectation and maximization (EM) framework. The extensive experimental evaluation shows that our bounding strategies allow IBIMiner to considerably reduce the runtime and the EM optimization can improve the predictive performance dramatically. We also show that our model can perform better than some of the state of the art regression models.
  • Thumbnail Image
    Item
    AREM: A Novel Associative Regression Model Based on EM Algorithm
    (2012-10-16) Jiang, Zhonghua; Karypis, George
    In recent years, there have been increasing efforts in applying association rule mining to build Associative Classification (AC) models. However, the similar area that applies association rule mining to build Associative Regression (AR) models has not been well explored. In this work, we fill this gap by presenting a novel regression model based on association rules called AREM. AREM starts with finding a set of regression rules by applying the instance based pruning strategy, in which the best rules for each instance are discovered and combined. Then a probabilistic model is trained by applying the EM algorithm, in which the right hand side of the rules and their importance weights are updated. The extensive experimental evaluation shows that our model can perform better than both the previously proposed AR model and some of the state of the art regression models, including Boosted Regression Trees, SVR, CART and Cubist, with the Mean Squared Error (MSE) being used as the performance metric.
  • Thumbnail Image
    Item
    Automatic Detection Of Vaccine Adverse Reactions By Incorporating Historical Medical Conditions
    (2011-03-21) Jiang, Zhonghua; Karypis, George
    Identifying medical conditions that are correlated with vaccine adverse reactions can not only provide better understanding of how adverse reactions are triggered but also have the potential of detecting new adverse reactions that are otherwise hidden. We formulate this problem as mining frequent patterns with constraints. The major constraint we use is called the minimum dual-lift constraint, where dual-lift is a novel measure we propose to evaluate correlations in a pattern. We also introduce the notation of minimum improvement constraint to remove redundancy in generated pattern set. We come up with a novel approach to upper bound the dual-lift measure which helps to prune the search space. Experimental results show that our algorithm works significantly better than the baseline on dense datasets. Our algorithm is also tested on the real world VAERS database. Some interesting vaccine adverse reactions identified are presented.
  • Thumbnail Image
    Item