Researchers crack optimal interventions for large-scale influence spread
A new linear programming method solves a previously NP-hard problem using only statistical network data.
Deep Dive
The authors developed an approximate solution for optimal interventions on the Linear Threshold Model (LTM) in large-scale networks. Their approach uses a local mean-field approximation and reformulates the problem as a linear program with infinite constraints, which they then approximate with finitely many constraints. Validated on real-world networks, the method was compared with optimal seeding and state-of-the-art algorithms for least-cost influence.
Key Points
- Solves a previously NP-hard optimal intervention problem on the Linear Threshold Model using a local mean-field approximation.
- Reformulates the problem as a linear program with infinite constraints, then approximates with finite constraints for computational tractability.
- Validated on real-world networks, outperforming optimal seeding and state-of-the-art least-cost influence algorithms.
Why It Matters
Enables scalable, cost-effective influence campaigns on massive networks without needing full network data.