Title: A Quadratic Programming Approach to the Graph Edit Distance Problem
Speaker: Horst Bunke
Subject: Algorithms and Data Structures. Graph Theory.
Area: Computer Science
Type of school: Online video lectures
School name: VideoLectures.Net
Link to original media: http://videolectures.net/
Country: other
Course language: English
Course media: Video
Course duration:
Contributor: Lennert Sloth
Comments:
In this paper we propose a quadratic programming approach to computing the edit distance of graphs. Whereas the standard edit distance is defined with respect to a minimum-cost edit path between graphs, we introduce the notion of fuzzy edit paths between graphs and provide a quadratic programming formulation for the minimization of fuzzy edit costs. Experiments on real-world graph data demonstrate that our proposed method is able to outperform the standard edit distance method in terms of recognition accuracy on two out of three data sets.

Comments Number of comments: 0

Leave a Comment



 



 



 



 



(maximum message length of 2000 characters)