Overcomplete Deep Subspace Clustering Networks

Jeya Maria Jose Valanarasu, Vishal M. Patel; Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2021, pp. 746-755

Abstract


Deep Subspace Clustering Networks (DSC) provide an efficient solution to the problem of unsupervised subspace clustering by using an undercomplete deep auto-encoder with a fully-connected layer to exploit the self expressiveness property. This method uses undercomplete representations of the input data which makes it not so robust and more dependent on pre-training. To overcome this, we propose a simple yet efficient alternative method - Overcomplete Deep Subspace Clustering Networks (ODSC) where we use overcomplete representations for subspace clustering. In our proposed method, we fuse the features from both undercomplete and overcomplete auto-encoder networks before passing them through the self-expressive layer thus enabling us to extract a more meaningful and robust representation of the input data for clustering. Experimental results on four benchmark datasets show the effectiveness of the proposed method over DSC and other clustering methods in terms of clustering error. Our method is also not as dependent as DSC is on where pre-training should be stopped to get the best performance and is also more robust to noise.

Related Material


[pdf] [arXiv]
[bibtex]
@InProceedings{Valanarasu_2021_WACV, author = {Valanarasu, Jeya Maria Jose and Patel, Vishal M.}, title = {Overcomplete Deep Subspace Clustering Networks}, booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)}, month = {January}, year = {2021}, pages = {746-755} }