Our recent research can be categorized into the following clusters:
Mechanism design (MD) provides a game theoretic framework to explore if
the given social choice function may be implemented as an equilibrium
outcome of an induced game. On the other hand, machine learning (ML) seeks
to learn the preferences or types of the agents through available data or
through intelligent exploration. Modern web applications require both ML
and MD for a complete solution. ML and MD are well investigated as
individual problems but not together; interesting research questions arise
when we try to handle them together. In this context, we are currently
exploring multi-armed bandit problems where the arms are controlled by
strategic agents. Immediate applications include Internet advertsing,
crowdsourcing, demanad management in smart grids, procurement, etc.
There is a rich variety of algorithmic and complexity theoretic problems
in the area of social choice theory in general and voting theory in
particular. The bag of problems we are investigating includes:
manipulability of voting rules; detecting manipulators; kernelization
complexity of manipulation problems in voting; sampling techniques for
voting problems; and frugal bribery problems.
We have been engaged with a variety of problems in social network analysis
where we are exploring the use of game theoretic and other
techniques for solving these problems better. The following problems
continue to engage our attention: influence maximization for viral
marketing; influence limitation for spread of virus and rumor; modeling
the spread of preferences in social networks; game theoretic models for
social network analysis problems; etc.
Ramasuri Narayanam and Y. Narahari.
A novel, decentralized, local information based algorithm for community detection in social networks. CSI Journal of Computing. Volume 2, Number 1-2, 2013, pp. 40-50.
Crowdsourcing has emerged as a popular means of getting work done through
an open call to a crowd. There are many challenges involved in setting up
successful crowdsourcing platforms. Using mechanism design and machine
learning, we are looking into research challenges such as learning the
qualities of crowd workers, ensuring an acceptable level of quality,
designing allocations and pricing mechanisms that arrest manipulations by
strategic crowd workers, etc. The canvas of applications includes
geosensing (for healthcare and public welfare projects), online education,
crowdfunding, procurement of services, etc.