A Comprehensive Review of Propagation Models in Complex Networks: From Deterministic to Deep Learning Approaches

Bin Wu,Sifu Luo,C. Steve Suh

2024-10-03

Abstract:Understanding propagation mechanisms in complex networks is essential for fields like epidemiology and multi-robot networks. This paper reviews various propagation models, from traditional deterministic frameworks to advanced data-driven and deep learning approaches. We differentiate between static and dynamic networks, noting that static models provide foundational insights, while dynamic models capture real-world temporal changes. Deterministic models like the SIR framework offer clear mathematical insights but often lack adaptability to randomness, whereas stochastic models enhance realism at the cost of interpretability. Behavior-based models focus on individual decision-making, demanding more computational resources. Data-driven approaches improve accuracy in nonlinear scenarios by adapting to evolving networks, using either traditional models or model-free machine learning techniques. We explore supervised and unsupervised learning methods, as well as reinforcement learning, which operates without predefined datasets. The application of graph neural networks (GNNs) is also discussed, highlighting their effectiveness in modeling propagation in complex networks. The paper underscores key applications and challenges associated with each model type, emphasizing the increasing importance of hybrid and machine learning-based solutions in contemporary network propagation issues.

Social and Information Networks

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to understand and model the propagation mechanisms in complex networks. Specifically, the author explores a series of propagation models from deterministic models to deep - learning - based methods, aiming to provide theoretical and technical support for different fields (such as epidemiology, social media, communication networks, and multi - robot systems, etc.).

### Core Problems of the Paper

1. **Understanding Propagation Mechanisms**:
- How information, viruses, or influences in complex networks propagate among nodes.
- The dynamic characteristics of the propagation process and its dependence on network structure and time changes.

2. **Improving Propagation Models**:
- From traditional deterministic models (such as the SIR model) to models that introduce randomness and are data - driven, in order to more accurately describe the propagation phenomena in the real world.
- Explore the respective advantages and disadvantages of different types of models (deterministic, stochastic, behavior - based, data - driven), and analyze their applicable scenarios.

3. **Applications and Challenges**:
- Research the performance of these models in practical applications, such as predicting and controlling the spread of epidemics, optimizing information dissemination strategies, etc.
- Emphasize the important role of hybrid models and machine - learning - based models in solving modern network propagation problems.

### Model Classification

1. **Deterministic Models**:
- For example, the classic SIR model, which describes the propagation process through differential equations.
- Formula representation:
\[
P(\text{Susceptible node } i \text{ being infected}) = 1 - \prod_{j \in N(i)} (1 - \beta A_{ij} I_j)
\]
- Advantage: Clear mathematical structure and easy to interpret.
- Disadvantage: Lack of flexibility and difficult to handle the randomness and uncertainty in the real world.

2. **Stochastic Models**:
- Introduce random variables to make the simulation more realistic, but increase the computational complexity and difficulty of interpretation.
- For example, the random walk model and the diffusion process.
- Formula representation:
\[
p_i(t + 1) = \sum_{j \in N(i)} P_{ji} p_j(t)
\]

3. **Behavior - Based Models**:
- Focus on the individual decision - making process, such as the agent - based model (ABM), which can capture complex individual behaviors.
- For example:
\[
S_i(t + 1) = f(S_i(t), \{S_j(t) | j \in N(i)\}, \epsilon_i(t))
\]

4. **Data - Driven Models**:
- Utilize big data and machine - learning methods to adapt to the ever - changing network environment and improve accuracy in nonlinear and dynamic scenarios.
- Include supervised learning, unsupervised learning, reinforcement learning, and graph neural networks (GNN), etc.

### Summary

This paper aims to provide researchers with a comprehensive perspective through a comprehensive review of various propagation models, helping them choose the most appropriate models and tools to solve the propagation problems in complex networks. Future research directions may include the integration and improvement of methods, as well as in - depth exploration in different application fields.