Transformed Subspace Clustering

Subspace clustering assumes that the data is separable into separate subspaces. Such a simple assumption, does not always hold. We assume that, even if raw subspaces, one can learn representation (transform coefficients) such learnt To achieve intended goal, we embed subspace techniques (locally linear manifold clustering, sparse and low rank representation) transform learning. The entire formulation jointly learnt; giving rise to new class of methods called transformed (TSC). In order account for non-linearity, kernelized extensions TSC are also proposed. test performance proposed techniques, benchmarking performed on image document datasets. Comparison with state-of-the-art shows our improves upon them.