Towards the Use of Metaheuristics for Optimizing the Combination of Classifier and Cluster Ensembles

Unsupervised models can provide supplementary soft constraints to help classify new data since similar instances are more likely to share the same class label. In this context, we investigate how to make an existing algorithm, named C3E (from Combining Classifier and Cluster Ensembles), more user-friendly by automatically tunning its main parameters with the use of metaheuristics. In particular, the C3E algorithm is based on a general optimization framework that takes as input class membership estimates from existing classifiers, as well as a similarity matrix from a cluster ensemble operating solely on the new (target) data to be classified, and yields a consensus labeling of the new data. To do so, two parameters have to be defined a priori, namely: the relative importance of classifier and cluster ensembles and the number of iterations of the algorithm. In some practical applications, these parameters can be optimized via (time consuming) grid search approaches based on cross-validation procedures. This paper shows that metaheuristics can be more computationally efficient alternatives for optimizing such parameters. More precisely, analyses of statistical significance made from experiments performed on fourteen datasets show that five metaheuristics can yield classifiers as accurate as those obtained from grid search, but taking half the running time.