Spectral Algorithms

Author(s):  
R. Kannan ◽  
S. Vempala
Keyword(s):  
Author(s):  
Mark Newman

The study of networks, including computer networks, social networks, and biological networks, has attracted enormous interest in recent years. The rise of the Internet and the wide availability of inexpensive computers have made it possible to gather and analyse network data on an unprecendented scale, and the development of new theoretical tools has allowed us to extract knowledge from networks of many different kinds. The study of networks is broadly interdisciplinary and developments have occurred in many fields, including mathematics, physics, computer and information sciences, biology, and the social science. This book brings together the most important breakthroughts in each of these fields and presents them in a unified fashion, highlighting the strong interconnections between work in different areas. Topics covered include the measurement of networks; methods for analysing network data, including methods developed in physics, statistics, and sociology; fundamentals of graph theory; computer algorithms, including spectral algorithms and community detection; mathematical models of networks such as random graph models and generative models; and models of processes taking place on networks.


2017 ◽  
Vol 5 (1) ◽  
pp. 139-157 ◽  
Author(s):  
Sam Cole ◽  
Shmuel Friedland ◽  
Lev Reyzin

Abstract In this paper, we consider the planted partition model, in which n = ks vertices of a random graph are partitioned into k “clusters,” each of size s. Edges between vertices in the same cluster and different clusters are included with constant probability p and q, respectively (where 0 ≤ q < p ≤ 1). We give an efficient algorithm that, with high probability, recovers the clusters as long as the cluster sizes are are least (√n). Informally, our algorithm constructs the projection operator onto the dominant k-dimensional eigenspace of the graph’s adjacency matrix and uses it to recover one cluster at a time. To our knowledge, our algorithm is the first purely spectral algorithm which runs in polynomial time and works even when s = Θ (√n), though there have been several non-spectral algorithms which accomplish this. Our algorithm is also among the simplest of these spectral algorithms, and its proof of correctness illustrates the usefulness of the Cauchy integral formula in this domain.


Author(s):  
Weiwei Du ◽  
◽  
Kiichi Urahama

We present unsupervised and semi-supervised algorithms for extracting fuzzy clusters in weighted undirected regular, undirected bipartite, and directed graphs. We derive the semi-supervised algorithms from the Lagrangian function in unsupervised methods for extracting dominant clusters in a graph. These algorithms are robust against noisy data and extract arbitrarily shaped clusters. We demonstrate applications for similarity searches of data such as image retrieval in face images represented by undirected graphs, quantized color images represented by undirected bipartite graphs, and Web page links represented by directed graphs.


Sign in / Sign up

Export Citation Format

Share Document