Faculty Contact: Jason Marden
Abstract: The maximization of submodular functions is a known NP-Hard problem for some classes of submodular functions. Consequently, many techniques exist to approximate the optimal solution. One such technique is a greedy algorithm, which can be implemented in a distributed way, using agents that make decisions sequentially.
This work aims to answer the following:
1. How does the communication structure (modeled as a graph) of the agents affect the accuracy of the algorithm?
2. What information is the most valuable to communicate?
3. What is the best communication structure?
- Fall 2017: David Grimsman