Learn More
We present a novel approach to automatically discover object categories from a collection of unlabeled images. This is achieved by the Information Bottleneck method, which finds the optimal partitioning of the image collection by maximally preserving the relevant information with respect to the latent semantic residing in the image contents. In this method,(More)
To solve the problem of determining the correct number of clusters, this paper proposes a new cluster validity index, IB_Hindex, for hierarchical clustering based on IB method. The index effectively incorporates the cluster cohesion and separation so that the corresponding algorithm is able to find the number of feature patterns hidden in dataset. IB_Hindex(More)
We present a novel unsupervised data analysis method, Multi-feature Information Bottleneck (MfIB), which is an extension of the Information Bottleneck (IB). In comparison with the original IB, the proposed MfIB method can analyze the data simultaneously from multiple feature variables , which characterize the data from multiple cues. To verify the(More)
Alternative clustering aims at exploring another reasonable clustering which is distinctively different from an existing one. This paper presents a novel alternative clustering algorithm based on the IB method, named Alt_sIB. Our approach aims to ensure the clustering quality by maximizing the mutual information between clustering labels and data(More)
  • 1