Events Calendar

YINS Summer Seminar: Soheil Eshghi & Marko Mitrovic

YINS Summer Seminar: Soheil Eshghi & Marko Mitrovic

Event time: 
Wednesday, July 26, 2017 - 12:00pm to 1:00pm
Location: 
Yale Institute for Network Science See map
17 Hillhouse Avenue, 3rd floor
New Haven, CT 06511
Event description: 

“Influence in social systems”

Speaker 1: Soheil Eshghi, Ph.D.

Talk summary: I will be presenting two pieces of ongoing work on effectively using a limited influence budget to achieve social goals.

In the first half, I will discuss optimal strategies for budget allocation to multiple influence channels across time for a political campaign seeking to win an election. We show that for a general set of objective functions, the optimal influence strategy is to exert maximum effort in waves for every channel, and then to cease effort and to let the effects propagate. We also show that early on, the total cost-adjusted reach of a channel determines its relative value, while targeting matters more closer to election time. Through our analyses, we identify a new and adaptable temporally varying  centrality metric, and show how it can effectively be used in the computation of these optimal allocations.

In the second half (time-permitting), I will discuss  a mathematical model for the stability of social groups under external influence. We define a notion of stability to be the minimum additional incentive that will motivate a group member to dissociate with a group, and describe this value for the most stable (fairness-constrained) redistributive norms. We show that with no fairness constraints, the most stable norms balance out the externalities of group membership, while under fairness constraints, they lead to the creation of two classes of group members, one of which is uniformly better off.

Bio: Soheil is a postdoctoral associate at YINS working on influence management with Leandros Tassiulas. He received his PhD from Penn and conducted postdoctoral research at Cornell. His research interests are in constrained decision-making for biological, technological, and social systems.

“Differentially Private Submodular Maximization”

Speaker 2: Marko Mitrovic

Talk summary: Modern machine learning has been wildly successful thanks in large part to the vast amount of data that is available in today’s world. However, at the same time, people are becoming increasingly worried that their personal information is being compromised. This talk will describe how we can use confidential data in various applications, while provably maintaining privacy. 
 
Bio: Marko Mitrovic is a 2nd year Ph.D. student in the Yale Computer Science Department, where he is advised by Amin Karbasi. His research interests are primarily centered around Machine Learning and Applied Graph Theory.
 
 
 
 
 
 
.