Slideshare uses cookies to improve functionality and performance, and to provide you with relevant advertising. If you continue browsing the site, you agree to the use of cookies on this website. See our User Agreement and Privacy Policy.

Slideshare uses cookies to improve functionality and performance, and to provide you with relevant advertising. If you continue browsing the site, you agree to the use of cookies on this website. See our Privacy Policy and User Agreement for details.

Successfully reported this slideshow.

Like this presentation? Why not share!

- Laplacian Colormaps: a framework fo... by Davide Eynard 4422 views
- icml2004 tutorial on spectral clust... by zukun 1314 views
- icml2004 tutorial on spectral clust... by zukun 1164 views
- Introduction to Sparse Methods by Shadi Nabil Albar... 981 views
- Distributed approximate spectral cl... by Bita Kazemi 483 views
- Spectral clustering Tutorial by Zitao Liu 7411 views

2,748 views

Published on

Published in:
Technology

No Downloads

Total views

2,748

On SlideShare

0

From Embeds

0

Number of Embeds

1,662

Shares

0

Downloads

45

Comments

0

Likes

2

No embeds

No notes for slide

- 1. Notes on Spectral Clustering Politecnico di Milano, 28/05/2012 Davide Eynard Institute of Computational Sciences - Faculty of Informatics Università della Svizzera Italiana davide.eynard@usi.ch
- 2. Talk outline Spectral Clustering Distances and similarity graphs Graph Laplacians and their properties Spectral clustering algorithms SC under the hood24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 2/15
- 3. Similarity graph The objective of a clustering algorithm is partitioning data into groups such that: Points in the same group are similar Points in different groups are dissimilar Similarity graph G=(V,E) (undirected graph) Vertices vi and vj are connected by a weighted edge if their similarity is above a given threshold GOAL: find a partition of the graph such that: edges within a group have high weights edges across different groups have low weights24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 3/15
- 4. Weighted adjacency matrix Let G(V,E) be an undirected graph with vertex set V={v1,...,vn} Weighted adjacency matrix W=(wij )i,j=1,...,n wij≥0 is the weight of the edge between vi and vj wij=0 means that vi and vj are not connected by an edge wij=wji Degree of a vertex vi∈V: di=∑j=1..nwij Degree matrix D=diag(d1,...,dn)24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 4/15
- 5. Different similarity graphs ε-neighborhood Connect all points whose pairwise distance is less than ε k-nearest neighbors if vi ∈ knn(vj) OR vj ∈ knn(vi) if vi ∈ knn(vj) AND vj ∈ knn(vi) (mutual knn) after connecting edges, use similarity as weight fully connected all points with similarity sij>0 are connected To control neighborhoods to be local, use a similarity function like the Gaussian: s(xi,xj)=exp(-║xi-xj║2/(2σ2))24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 5/15
- 6. Graph Laplacians Graph Laplacian: L=D–W (symmetric and positive semi-definite) Properties Smallest eigenvalue λ1=0 with eigenvector = � n non-negative, real-valued eigenvalues 0=λ1≤λ2≤...≤λn the multiplicty k of the eigenvalue 0 of L equals the number of connected components A1,...,Ak in the graph24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 6/15
- 7. Spectral Clustering algorithm (1) Spectral Clustering algorithm Input: Similarity matrix S ∈ ℝn×n, number k of clusters to construct. 1. Construct a similarity graph as previously described. Let W be its weighted adjacency matrix. 2. Compute the unnormalized Laplacian L 3. Compute the first k eigenvectors u1,…,uk of L 4. Let U ∈ ℝn×k be the matrix containing the vectors u1,…,uk as columns 5. For i=1,...,n let yi ∈ ℝk be the vector corresponding to the i-th row of U 6. Cluster the points (yi)i=1,...,n in ℝk with the k-means algorithm into clusters C1,...,Ck. Output: Clusters A1,...,Ak with Ai= {j | yj ∈ Ci}.24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 7/15
- 8. Normalized Graph Laplacians Normalized graph Laplacians Symmetric: Lsym=D-1/2LD-1/2 = I-D-1/2WD-1/2 Random Walk: Lrw=D-1L=I-D-1W Properties λ is an eigenvalue of Lrw with eigenvector u iff λ is an eigenvalue of Lsym with eigenvector w=D1/2u λ is an eigenvalue of Lrw with eigenvector u iff λ and u solve the generalized eigenproblem Lu=λDu24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 8/15
- 9. Normalized Graph Laplacians Normalized graph Laplacians Symmetric: Lsym=D-1/2LD-1/2 = I-D-1/2WD-1/2 Random Walk: Lrw=D-1L=I-D-1W Properties (follow) 0 is an eigenvalue of Lrw with � as eigenvector, and an eigenvalue of Lsym with eigenvector D1/2�. Lsym and Lrw are positive semi-definite and have n non- negative, real-valued eigenvalues 0=λ1≤λ2≤...≤λn the multiplicty k of the eigenvalue 0 of both Lsym and Lrw equals the number of connected components A1,...,Ak24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 9/15
- 10. Spectral Clustering algorithm (2) Normalized Spectral Clustering Input: Similarity matrix S ∈ ℝn×n, number k of clusters to construct. Lrw: 3. Compute the first k generalized eigenvectors u1,…,uk of the generalized eigenproblem Lu=λDu Lsym: 2. Compute the normalized Laplacian Lsym 3. Compute the first k eigenvectors u1,…,uk of Lsym 4. normalize the eigenvectors Output: Clusters A1,...,Ak with Ai= {j | yj ∈ Ci}.24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 10/15
- 11. A spectral clustering example24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 11/15
- 12. Under the hood 0-eigenvalues in the ideal case parameters are crucial: k in k nearest neighbors σ in Gaussian kernel k (another one!) in k-means24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 12/15
- 13. Random Walk point of view Random walk: stochastic process which randomly jumps from one vertex to another Clustering: finding a partition such that a random walk stays long within a cluster and seldom jumps between clusters Transition probability pij:=wij/di Transition matrix: P = D-1W => Lrw= I - P λ is an eigenvalue of Lrw with eigenvector u iff 1-λ is an eigenvalue of P with eigenvector u24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 13/15
- 14. References Von Luxburg, U. (2007). A tutorial on spectral clustering. Statistics and Computing, 17(4), 395-416. Springer.24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 14/15
- 15. Thank you! Thanks for your attention! Questions?24/04/2011 Notes on Spectral Clustering and Joint Diagonalization 15/15

No public clipboards found for this slide

Be the first to comment