2012/2013
Clustering with uncertainties: An affinity propagation-based approach
The 19th International Conference on Neural Information Processing (ICONIP 2012), Doha, Qatar
| Author(s) | Wenye Li |
|---|---|
| Summary | Clustering is a classical unsupervised learning technique which has wide applications. One popular clustering model seeks a set of centers and organizes the data into different groups, with an objective to maximize the net similarities within each cluster. In this paper, we first formulate a generalized form of the clustering model, where the similarity measure has uncertainties or changes in different states. Then we propose an affinity propagation-based algorithm, which gives an efficient and accurate solution to the generalized model. Finally we evaluate the model and the algorithm by experiments. The results have justified the usefulness of the model and demonstrate the improvements of the algorithm over other possible solutions. |