Boosting a weak learning algorithm by majority
Information and Computation
Machine Learning
A decision-theoretic generalization of on-line learning and an application to boosting
Journal of Computer and System Sciences - Special issue: 26th annual ACM symposium on the theory of computing & STOC'94, May 23–25, 1994, and second annual Europe an conference on computational learning theory (EuroCOLT'95), March 13–15, 1995
On the Optimality of the Simple Bayesian Classifier under Zero-One Loss
Machine Learning - Special issue on learning with probabilistic representations
Class Decomposition via Clustering: A New Framework for Low-Variance Classifiers
ICDM '03 Proceedings of the Third IEEE International Conference on Data Mining
Combining Pattern Classifiers: Methods and Algorithms
Combining Pattern Classifiers: Methods and Algorithms
Data Mining: Practical Machine Learning Tools and Techniques, Second Edition (Morgan Kaufmann Series in Data Management Systems)
Classifier Ensembles with a Random Linear Oracle
IEEE Transactions on Knowledge and Data Engineering
Naïve Bayes ensembles with a random oracle
MCS'07 Proceedings of the 7th international conference on Multiple classifier systems
Using supervised clustering to enhance classifiers
ISMIS'05 Proceedings of the 15th international conference on Foundations of Intelligent Systems
A survey: hybrid evolutionary algorithms for cluster analysis
Artificial Intelligence Review
Hi-index | 0.00 |
Classical clustering algorithms require a predefined number of cluster centers. They are often very sensitive to initialization, which can result in very different clustering results. We present a two-phase algorithm which is a combination of a new ant based algorithm and a nonnegative matrix factorization-based consensus clustering algorithm. Ant clustering approaches can and do find the number of clusters as well as the data partition. However, they are very sensitive to both initial conditions and select parameters. Here, we show that using an ensemble of ant partitions and NMF to combine them we can find both the "right" number of clusters and a good data partition. Experiments were done with ten data sets. We conducted a wide range of comparisons that demonstrate the effectiveness of this new approach.