Elliot Anshelevich

My research interests center in the design and analysis of algorithms, especially in large decentralized networks. My main focus has largely been on algorithmic game theory, and more specifically on networks involving strategic agents. In order to understand such networks, I also study diffusion processes in social networks, and more generally approximation algorithms. An increasing number of networks at the forefront of scientific research, and of great importance to our world, consist of interactions of a large number of independent agents that we cannot control. Contrary to this, traditional study of network design assumes that there is a single network designer, and that its design is accepted and implemented by all parts of the network. Similarly, study involving network management assumes that some protocol is implemented in the network, and all parts of the network follow this protocol (or almost all parts, while a small part of the network is malicious). The lack of control of the agents can occur because of the self-interest of these independent agents (such as in peer-to-peer file sharing schemes, business contracts between companies, transportation networks consisting of individual cars, or social networks representing relationships between groups of people), or because the process involved inherently lacks control by outside forces (such as in epidemics spreading through a population). The outcomes and dynamics of such behavior often have very different properties from those of the centrally designed or managed networks. Therefore, dealing with such systems requires quite different methods and considerations, leading to new and exciting algorithmic questions. This fact and the enormous growth in the complexity of networked systems have made game theoretic techniques and networks of self-interested agents become a common theme in computer science. The major goal of my research is to design algorithms and develop mechanisms for dealing with and understanding such networks. While these networks cannot be fully controlled, they can often be influenced in a limited way, sometimes resulting in dramatic improvement of the global network behavior. I have focused on algorithms determining how such influence can be applied, as well as studying agent dynamics, and understanding the properties of equilibrium outcomes formed by such agents. The emphasis of my work is usually on algorithms and network properties with provable guarantees. My current major research topics are described below.