Chapter14. Influence Maximization in Networks
투빅스 13기 이예지
Viral Marketing



Two Classical Propagation Models
Linear Threshold Model
Independent Cascade Model
How hard is influence maximization?
Hill Climbing(Greedy)

의 두 가지 properties

Plan: Prove 2 things (1) Our f(S) is submodular



Principle of deferred decision

Plan: Prove 2 things (2) Hill Climbing gives near optimal solutions



Speeding things up: Sketch-based Algorithms


keep multiple ranks


Experiments

PreviousChapter13. Probabilistic Contagion and Models of influnceNextChapter15. Outbreak Detection in Networks
Last updated