Future high-performance networks will rely on smart devices able to cooperate, take decisions in a distributed fashion according to their own objectives, reconfigure dynamically by reacting to the envi- ronment. Such behavior requires efficient techniques to coordinate the actions of different nodes who are not necessarily obedient but rather strategic. Thus, a key challenge is to understand the interaction between two different types of constraints: the strategic constraints, imposed by the decentralized nature of the decision process, and the information processing contraints, imposed by the environment such as noise effect, network’s topology or exogenous restrictions on the number of messages.
The PhD candidate will address the problem of “strategic communication” by leveraging recent re- sults [1, 2] at the interplay between Information Theory and Game Theory. The key idea is to model the actions performed by the nodes by discrete random variables, and to measure the level of coordina- tion [3, 4, 5] by the distance between their joint distribution and a target distribution. In contrast to the information-theoretic paradigm [7], we assume that the nodes optimally disclose their information according to their own objectives, e.g. as in the “Bayesian persuasion game” [6].
The highly motivated PhD applicant must have good mathematical skills. The PhD thesis will be supervised by Maël Le Treust, CNRS permanent researcher. Doctoral contract is funded by the Paris Seine Initiative and offers a remuneration of 40k e /year for a duration of three years. Applicant should send a short email together with master’s degree transcripts, recommandation letters and Curriculum Vitæ to the address below.
Expected start date: October - December 2018 (flexible).
Location: Laboratory ETIS UMR 8051, Université Paris Seine, Université Cergy-Pontoise, ENSEA, CNRS, 6 avenue du Ponceau, 95014 Cergy-Pontoise, France.
References
[1] M. Le Treust and T. Tomala, “Information-theoretic limits of strategic communication,” draft available: https://arxiv.org/abs/1807.05147, July 2018.
[2] M. Le Treust and T. Tomala, “Persuasion with limited communication capacity,” draft available: https://arxiv.org/abs/1711.04474, Dec. 2017.
[3] O. Gossner, P. Hernàndez, and A. Neyman, “Optimal use of communication resources,” Econometrica, vol. 74, no. 6, pp. 1603–1636, 2006.
[4] P. Cuff, H. Permuter, and T. Cover, “Coordination capacity,” 鶹ýӳ Transactions on Information Theory, vol. 56, no. 9, pp. 4181–4206, 2010.
[5] M. Le Treust, “Joint empirical coordination of source and channel,” 鶹ýӳ Transactions on Information Theory, vol. 63, pp. 5087–5114, Aug 2017.
[6] E. Kamenica and M. Gentzkow, “Bayesian persuasion,” American Economic Review, vol. 101, pp. 2590–2615, 2011.
[7] A. E. Gamal and Y.-H. Kim, Network Information Theory. Cambridge University Press, Dec. 2011.
Contact: Maël Le Treust ([email protected])
Phone: +33 (0)1 30 73 62 64
Webpage: https://sites.google.com/site/maelletreust/