DocumentCode :
2505676
Title :
Learnability of latent position network models
Author :
Choi, David S. ; Wolfe, Patrick J.
Author_Institution :
Sch. of Eng. & Appl. Sci., Harvard Univ., Cambridge, MA, USA
fYear :
2011
fDate :
28-30 June 2011
Firstpage :
521
Lastpage :
524
Abstract :
The latent position model is a well known model for social network analysis which has also found application in other fields, such as analysis of marketing and e-commerce data. In such applications, the data sets are increasingly massive and only partially observed, giving rise to the possibility of overfitting by the model. Using tools from statistical learning theory, we bound the VC dimension of the latent position model, leading to bounds on the overfit of the model. We find that the overfit can decay to zero with increasing network size even if only a vanishing fraction of the total network is observed. However, the amount of observed data on a per-node basis should increase with the size of the graph.
Keywords :
graph theory; learning (artificial intelligence); social networking (online); statistical analysis; e-commerce data; graph size; latent position network model; marketing data; network model learnability; social network analysis; statistical learning theory; Analytical models; Atmospheric modeling; Biological system modeling; Bipartite graph; Data models; Social network services; Upper bound; extremal graph theory; latent position model; learning theory; random graphs; social network analysis;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Statistical Signal Processing Workshop (SSP), 2011 IEEE
Conference_Location :
Nice
ISSN :
pending
Print_ISBN :
978-1-4577-0569-4
Type :
conf
DOI :
10.1109/SSP.2011.5967748
Filename :
5967748
Link To Document :
بازگشت