Learning hidden cascades via classification

2025-05-16

AI Summary

In real-world social networks, individual propagation states (e.g., infected/informed) are often latent, with only intermediate observable signals (e.g., symptoms) available—rendering conventional epidemic modeling ineffective. This paper introduces “distributional classification,” a novel paradigm that, for the first time, leverages classifier generalization capacity to learn latent-state dynamics, thereby relaxing the full-observability assumption. Our method constructs a distribution-level classification framework via machine learning and validates it on synthetic network simulations and a real insider-trading network. It significantly improves accuracy in inferring transmission paths and outbreak sources on highly cyclic, densely connected complex networks. Key contributions include: (1) establishing the learnability-theoretic foundation for latent-state propagation models; (2) enabling end-to-end dynamical inference solely from observable intermediate signals; and (3) outperforming state-of-the-art baselines under strong state concealment.

Technology Category

• Machine Learning: Probabilistic Circuits and Graphical Models
• Reasoning under Uncertainty: Relational Probabilistic Models
• Search and Optimization: Distributed Search

Application Category

• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

The spreading dynamics in social networks are often studied under the assumption that individuals' statuses, whether informed or infected, are fully observable. However, in many real-world situations, such statuses remain unobservable, which is crucial for determining an individual's potential to further spread the infection. While this final status is hidden, intermediate indicators such as symptoms of infection are observable and provide important insights into the spread process. We propose a partial observability-aware Machine Learning framework to learn the characteristics of the spreading model. We term the method Distribution Classification, which utilizes the power of classifiers to infer the underlying transmission dynamics. We evaluate our method on two types of synthetic networks and extend the study to a real-world insider trading network. Results show that the method performs well, especially on complex networks with high cyclic connectivity, supporting its utility in analyzing real-world spreading phenomena where direct observation of individual statuses is not possible.

Problem

• Estimating hidden spreading dynamics in networks with partial observability
• Inferring transmission models using observable intermediate indicators
• Analyzing real-world networks where individual statuses are unobservable

Innovation

• Partial observability-aware Machine Learning framework
• Distribution Classification for transmission dynamics
• Effective on high cyclic connectivity networks