We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Network Diffusions via Neural Mean-Field Dynamics

Abstract: We propose a novel learning framework based on neural mean-field dynamics for inference and estimation problems of diffusion on networks. Our new framework is derived from the Mori-Zwanzig formalism to obtain an exact evolution of the node infection probabilities, which renders a delay differential equation with memory integral approximated by learnable time convolution operators, resulting in a highly structured and interpretable RNN. Directly using cascade data, our framework can jointly learn the structure of the diffusion network and the evolution of infection probabilities, which are cornerstone to important downstream applications such as influence maximization. Connections between parameter learning and optimal control are also established. Empirical study shows that our approach is versatile and robust to variations of the underlying diffusion network models, and significantly outperform existing approaches in accuracy and efficiency on both synthetic and real-world data.
Comments: Accepted by NIPS2020, 21 pages, 5 figures
Subjects: Machine Learning (cs.LG); Social and Information Networks (cs.SI); Machine Learning (stat.ML)
Cite as: arXiv:2006.09449 [cs.LG]
  (or arXiv:2006.09449v3 [cs.LG] for this version)

Submission history

From: Shushan He [view email]
[v1] Tue, 16 Jun 2020 18:45:20 GMT (417kb,D)
[v2] Thu, 22 Oct 2020 20:41:16 GMT (537kb,D)
[v3] Tue, 19 Jan 2021 16:15:10 GMT (1062kb,D)

Link back to: arXiv, form interface, contact.