Fig. 11From: A novel algorithm for fast and scalable subspace clustering of high-dimensional dataNo of found subspaces/clusters vs runtime (madelon dataset). Different epsilon values ranging from 1.0E−5 to 1.0E−6 were used. The number of clusters as well as subspaces in which these clusters are found, increases with the increase in ε value. Other parameters used: τ=3,m inSize=4Back to article page