Sparse Subspace Clustering (SSC)

We looked into sparse subspace clustering (SSC). SSC was introduced by Elhamifar and Vidal in their paper “Sparse Subspace Clustering: Algorithm, Theory, and Applications”. It was revisited by Matsushima and Brbic in their paper “Selective Sampling-based Scalable Sparse Subspace Clustering”. In this video, the focus is on (1) how to find a sparse solution for subspace clustering (2) the difference between independent and disjoint subspaces (3) coding SSC in python and testing it on a couple of synthetic datasets.
🔗 jupyter notebook I built in this video:
Spectral clustering tutorial:
– my website ➡️
– my github ➡️
– my linkedin ➡️
📹 Video edit: Adobe Premiere Rush
🎧 Audio enhancement: Adobe Podcast
0:00 start
0:13 reading the paper
0:52 problem statement
1:39 an example of a subspace
2:12 self-expressiveness property
3:32 visualizing sparse-subspace representation
5:46 my notes on representing a point using other points from the subspace
6:26 independent and disjoint subspaces
7:55 sparse subspace clustering (SSC) algorithm
8:12 code for sparse subspace clustering (SSC)
8:40 creating synthetic datasets
10:03 a function for finding a sparse solution
11:43 a function for SSC adjacency
12:52 my notes on numpy argpartition
14:13 visualizing the adjacency matrix
15:13 the results of clustering
16:13 clustering iris dataset
17:11 final remarks
#graph #spectralclustering #subspace #subspaceclustering #sparsesubspaceclustering #SSC #machinelearning #python #numpy #ai #artificialintelligence #PCA

Duration: 00:18:06