Difference between revisions of "Melia et al AISTATS 2001"
From Cohen Courses
Jump to navigationJump to search| Line 9: | Line 9: | ||
== Summary == | == Summary == | ||
| − | This paper gives a general theoretical interpretation of a wide variety of spectral methods. The authors first present the general framework of normalized cuts. We assume we are given an index set <math>I</math> and similarity matrix <math>S</math> where entry <math>S_{ij}</math> represents a similarity between item <math>i</math> and item <math>j</math>. We then seek a partition of <math>I</math> into two sets <math>A,\overline{A}</math>. | + | This paper gives a general theoretical interpretation of a wide variety of spectral methods. The authors first present the general framework of normalized cuts. We assume we are given an index set <math>I</math> and similarity matrix <math>S</math> where entry <math>S_{ij}</math> represents a similarity between item <math>i</math> and item <math>j</math>. We then seek a partition of <math>I</math> into two sets <math>A,\overline{A}</math>. Minimizing the following criteria: |
| + | <math> | ||
| + | NCut(A, \overline{A}) = \left( \frac{1}{Vol(A)}\frac{1}{vol(\overline{A})}\right) | ||
| + | <math> | ||
== Related papers == | == Related papers == | ||
Revision as of 13:55, 4 February 2011
Citation
Marina Melia and Jianbo Shi. 2001. A Random Walks View of Spectral Segmentation. In AISTATS 2001.
Online version
Available on Marina Melia's Website
Summary
This paper gives a general theoretical interpretation of a wide variety of spectral methods. The authors first present the general framework of normalized cuts. We assume we are given an index set and similarity matrix where entry represents a similarity between item and item . We then seek a partition of into two sets . Minimizing the following criteria: <math> NCut(A, \overline{A}) = \left( \frac{1}{Vol(A)}\frac{1}{vol(\overline{A})}\right) <math>
