Learning and Inferring Networks from Incomplete Data
Abstract
The objective of this proposed research is to learn the characteristics of explicit and implicit networks, such as social networks using both observables and by responses to probes. Characteristics of the network to be learned consist both of explicit structure of hidden portions of the network as well as network characteristics and responses. The proposed research is divided into three main thrusts using both passive and active learning of the network: 1. Passive inference from constraints: Find the most likely network topology given a priori information, network data, and constraints using optimization and specialized data structures. 2. Passive inference from votes: Inferring network connections and other characteristics from formal or informal voting (or opinion) data. 3. Active inference in independent cascade models: Similar to the above, but actively probing the network and observing the resulting cascade. This would include design of the probe to maximize information with the minimum probing.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 12, 2017
- Source ID
- W911NF1510546
Entities
People
- Lev Reyzin
Organizations
- Army Contracting Command
- United States Army
- University of Illinois at Chicago