	Log in \| Register \|

Machine Learning seminars at the Cambridge University Engineering Department

Spectral Clustering

author: Arik Azran, Department of Engineering, University of Cambridge

Description

Machine Learning Tutorial Lecture

Spectral clustering is a technique for finding group structure in data. It is based on viewing the data points as nodes of a connected graph and clusters are found by partitioning this graph, based on its spectral decomposition, into subgraphs that posses some desirable properties.

My plan for this talk is to give a review of the main spectral clustering algorithms, demonstrate their abilities and limitations and offer some insight into when the method can be expected to be successful. No previous knowledge is assumed, and anyone who is interested in clustering (or fun applications of linear algebra) might find this talk interesting.

You might be experiencing some problems with Your Video player.

*Slides*
0:00	A Tutorial on Spectral Clustering
0:06	Good clustering – we know it when we see it
1:14	What is and what is not in the talk
3:22	Plan - Graph partitioning
4:06	Some Graph Terminology
6:15	Graph Cuts
9:20	Plan - Algorithms
9:23	Spectral clustering -overview
11:32	On spectral Clustering: Analysis and an algorithm
11:36	Spectral clustering -overview
16:15	On spectral Clustering: Analysis and an algorithm
17:00	Spectral clustering -overview
17:04	On spectral Clustering: Analysis and an algorithm
17:14	Spectral clustering -overview
17:20	On spectral Clustering: Analysis and an algorithm
20:49	Informal discussion: the ‘ideal’case (1)
21:14	Spectral clustering -overview
21:19	Informal discussion: the ‘ideal’case (1)
22:01	On spectral Clustering: Analysis and an algorithm
22:03	Informal discussion: the ‘ideal’case (1)
22:22	Informal discussion: the ‘ideal’case (2)
22:55	Informal discussion: the ‘ideal’case (1)
22:57	Informal discussion: the ‘ideal’case (2)
23:17	Informal discussion: the ‘ideal’case (1)
23:21	On spectral Clustering: Analysis and an algorithm
23:31	Informal discussion: the ‘ideal’case (1)
23:31	Informal discussion: the ‘ideal’case (2)
25:16	On spectral Clustering: Analysis and an algorithm
25:34	Informal discussion: the ‘ideal’case (2)
26:31	Informal discussion: the ‘ideal’case (1)
26:35	Informal discussion: the ‘ideal’case (2)
26:53	On spectral Clustering: Analysis and an algorithm
27:15	Spectral clustering -overview
27:17	On spectral Clustering: Analysis and an algorithm
27:18	Informal discussion: the ‘ideal’case (1)
27:20	Informal discussion: the ‘ideal’case (2)
27:32	Informal discussion: the ‘ideal’case (1)
28:49	On spectral Clustering: Analysis and an algorithm
28:52	Informal discussion: the ‘ideal’case (2)
29:31	Informal discussion: the ‘ideal’case (1)
30:00	On spectral Clustering: Analysis and an algorithm
30:20	Informal discussion: the ‘ideal’case (1)
31:05	Informal discussion: the ‘ideal’case (2)
31:06	A Random Walks View spectral Segmentation
32:37	On spectral Clustering: Analysis and an algorithm
32:47	A Random Walks View spectral Segmentation
35:39	On spectral Clustering: Analysis and an algorithm
36:34	Informal discussion: the ‘ideal’case (2)
36:55	A Random Walks View spectral Segmentation
37:08	Normalized Cuts and Image Segmentation
38:59	Connections
42:37	A Random Walks View spectral Segmentation
42:50	Normalized Cuts and Image Segmentation
42:53	Connections
43:12	Plan - Possible explanations
43:15	Possible explanations
44:23	Graph Laplacianand Smooth Functions
46:21	A Random Walks View spectral Segmentation
46:42	Graph Laplacianand Smooth Functions
49:17	Random Walks
49:23	Graph Laplacianand Smooth Functions

Lecture rating

People found this lecture:
Worth seeing
because it is:
Valuable and informative
Well presented
Easily understandable
Acceptably recorded
You need to login to cast your vote.

Report a problem or upload files

If you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.