CHAPTER 5

Nonparametric Mechanisms: Multiple Tasks

Another strategy for obtaining nonparametric mechanisms is to learn the parameters from the data submitted by the agents themselves, during some fixed time interval. This works whenever we have agents providing data about multiple (and ideally many) very similar phenomena within a short time interval, as shown in Figure 5.1.

Image

Figure 5.1: Scenario for multi-task mechanisms.

5.1  CORRELATED AGREEMENT

For example, we can observe the frequency of different submitted reports, and use this information to make the expected reward of any random reporting strategy equal to zero. Dasgupta and Ghosh [

Get Game Theory for Data Science now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.