Abstract:
This paper introduces an unsupervised feature learning architecture based on a multi-clustering integration module and MIRBM, a variant of RBM. The multi-clustering integration module uses three clusterers—K-means, affinity propagation, and spectral—to generate three clustering partitions (CPs) without background knowledge or labels. An unanimous voting strategy creates a local clustering partition (LCP). The proposed unsupervised feature learning architecture relies on the novel MIRBM model. Its novelty is that LCP as unsupervised guidance is integrated into one-step contrastive divergence (CD1) learning to guide hidden layer feature distribution. In the proposed architecture, the hidden and reconstructed hidden layer features of the MIRBM model in the same LCP cluster tend to constrict together during training. During training, LCP centers try to spread out in the hidden and reconstructed hidden layer. The proposed unsupervised feature learning architecture outperforms state-of-the-art clustering models in the Microsoft Research Asia Multimedia (MSRA-MM)2.0 dataset in feature representation and generalization.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
