Subspace clustering for high dimensional data: a review
ACM SIGKDD Explorations Newsletter - Special issue on learning from imbalanced datasets
Approximating the Knee of an MOP with Stochastic Search Algorithms
Proceedings of the 10th international conference on Parallel Problem Solving from Nature: PPSN X
| Hi-index | 0.00 |
Subspace clustering coevolves the attribute space supporting clusters at the same time as parameterizing the cluster location and combination. Typically, a 'flat' representation is pursued in which individuals describe both the property of individual clusters as well as the combination of clusters used to define the overall solution; hereafter F-ESC. Conversely, a symbiotic approach was recently proposed in which candidate clusters and the combination of clusters are coevolved from independent populations; hereafter S-ESC. In this work a common framework is pursued in order for flat and symbiotic evolutionary subspace clustering to be compared directly. We show that F-ESC might match S-ESC results for data sets with high proportions of cluster support, however, the gap between the two algorithm increases as cluster support decreases.