Learning Individual Behavior in Agent-Based Models with Graph Diffusion Networks

2025-05-27

AI Summary

To address the challenge of jointly optimizing non-differentiable Agent-Based Models (ABMs) with real-world data, this paper introduces the first end-to-end differentiable ABM framework. Unlike conventional surrogate modeling approaches—which approximate only system-level outputs—our method innovatively learns local, decentralized behavioral rules for each agent, thereby fully preserving the bottom-up dynamical essence of ABMs. Technically, we integrate graph diffusion models—to capture stochasticity in agent behavior—with graph neural networks—to encode multi-agent interaction structures—enabling joint modeling of individual agent trajectories and emergent population-level patterns. Evaluated on the Schelling segregation and predator–prey models, our framework reduces extrapolation prediction error by 42% compared to state-of-the-art surrogate methods, demonstrating substantial improvements in both fidelity and generalization.

Technology Category

• Multiagent Systems: Agent-Based Simulation and Emergent Behavior
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Agent-Based Models (ABMs) are powerful tools for studying emergent properties in complex systems. In ABMs, agent behaviors are governed by local interactions and stochastic rules. However, these rules are, in general, non-differentiable, limiting the use of gradient-based methods for optimization, and thus integration with real-world data. We propose a novel framework to learn a differentiable surrogate of any ABM by observing its generated data. Our method combines diffusion models to capture behavioral stochasticity and graph neural networks to model agent interactions. Distinct from prior surrogate approaches, our method introduces a fundamental shift: rather than approximating system-level outputs, it models individual agent behavior directly, preserving the decentralized, bottom-up dynamics that define ABMs. We validate our approach on two ABMs (Schelling's segregation model and a Predator-Prey ecosystem) showing that it replicates individual-level patterns and accurately forecasts emergent dynamics beyond training. Our results demonstrate the potential of combining diffusion models and graph learning for data-driven ABM simulation.

Problem

• Learning non-differentiable agent behavior rules in ABMs
• Modeling individual agent interactions using graph networks
• Forecasting emergent dynamics from decentralized agent behaviors

Innovation

• Differentiable surrogate for ABMs using observed data
• Combines diffusion models and graph neural networks
• Models individual agent behavior directly