This repository holds the code for paper Learnability of Competitive Threshold Models, IJCAI 2022.
Modeling the spread of social contagions is central to various applications in social computing. In this paper, we study the learnability of the competitive threshold model from a theoretical perspective. We simulate the competitive threshold models by artificial neural networks with finite VC dimensions, which enables analytical sample complexity and generalization bounds. Based on the proposed hypothesis space, we design efficient algorithms under the empirical risk minimization scheme.
python >= 3.6NumpyPyTorch >= 1.7sklearncplexdocplexnetworkx
The folder 'data' contains the synthetic and real datasets.
- Synthetic_data: Kronecker graph edge list file.
- real_data: retweet network edgelist file derived from Twitter.