Learning in social and economic networks
Abstract/Contents
- Abstract
- Social and economic networks are becoming increasingly important, both on the internet and otherwise. Agents in these networks possess limited information, and interact chiefly with their local neighborhood. Yet these networks have proven remarkably effective at the aggregation of 'information' at massive scales. It is of great scientific and commercial interest to build realistic models for phenomena in networks of agents. This thesis focuses on understanding the dynamics of interaction between agents on social and economic networks, and developing general theoretical tools in this context. Exchange networks model the behavior of a set of players who need to reach pairwise agreements for mutual benefit, as in the labor market, the housing market and the 'market' for social relationships. A crucial but little understood aspect of exchange networks is the dynamics of bargaining between players. We present a natural model of the bargaining dynamics in general networks, and show rapid convergence to certain socially optimal outcomes. We also describe internet-based experiments on bargaining in networks, which are the largest such experiments to date. In many contexts, agents 'learn' behavior from interaction with friends/neighbors on a network. We call this phenomenon 'social learning'. We focus on models of repeated interaction, with agents 'voting' in a series of rounds on some issue of interest. Votes in the initial round are based on 'private signals', whereas votes in future rounds incorporate knowledge of previous votes cast by friends. Thus, information is aggregated by the network via iterative updates. We consider two different models of iterative learning. A very simple model is 'majority dynamics' where agents choose their vote based on the majority of neighbors' votes in the previous round. We analyze this model on regular trees. At the other extreme is iterative Bayesian learning: a fully rational model introduced by Gale and Kariv (2003). We introduce new algorithms for this model, challenging a widespread belief that it is computationally intractable. We develop a novel technique -- the 'dynamic cavity method', which serves as a key analytical and algorithmic tool, and is generally applicable to social learning dynamics on tree networks. Finally, we prove fundamental limits on the rate of information aggregation in tree networks via social learning, when votes are discrete.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2012 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Kanoria, Yashodhan |
|---|---|
| Associated with | Stanford University, Department of Electrical Engineering |
| Primary advisor | Montanari, Andrea |
| Thesis advisor | Montanari, Andrea |
| Thesis advisor | El Gamal, Abbas A |
| Thesis advisor | Goel, Ashish |
| Advisor | El Gamal, Abbas A |
| Advisor | Goel, Ashish |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Yashodhan Kanoria. |
|---|---|
| Note | Submitted to the Department of Electrical Engineering. |
| Thesis | Thesis (Ph.D.)--Stanford University, 2012. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2012 by Yashodhan Kanoria
Also listed in
Loading usage metrics...