Learning models in social networks
This thesis studies the problem of modeling learning through a set of agents connected via a social network. We first analyze a robust detection problem in a tandem network where agents sequentially receive private signals about the state of the world, as well as the decision of their predecesso...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Theses and Dissertations |
| Language: | English |
| Published: |
2019
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/85449 http://hdl.handle.net/10220/50121 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | This thesis studies the problem of modeling learning through a set of agents connected via a social network.
We first analyze a robust detection problem in a tandem network where agents sequentially receive private signals about the state of the world, as well as the decision of their predecessors, and then attempt to make a decision about the state of the world. The knowledge they have regarding their predecessors' private signals, however, is incomplete, and we propose a policy to allow agents to minimize the worst-case error probability over all possible distributions of their predecessors' private signals.
We then consider the problem of online advertising, where an online retailer wishes to learn more about users in a social network. We model the problem in a variety of different ways. One of these models is optimizing a string-submodular function with incomplete information about the step-wise gains of each user. Another is through two different multi-armed bandit models. |
|---|