Deep learning reveals hidden interactions in complex systems

Abstract

Rich phenomena from complex systems have long intrigued researchers, and yet modeling system micro-dynamics and inferring the forms of interaction remain challenging for conventional data-driven approaches, being generally established by human scientists. In this study, we propose AgentNet, a model-free data-driven framework consisting of deep neural networks to reveal and analyze the hidden interactions in complex systems from observed data alone. AgentNet utilizes a graph attention network with novel variable-wise attention to model the interaction between individual agents, and employs various encoders and decoders that can be selectively applied to any desired system. Our model successfully captured a wide variety of simulated complex systems, namely cellular automata (discrete), the Vicsek model (continuous), and active Ornstein--Uhlenbeck particles (non-Markovian) in which, notably, AgentNet's visualized attention values coincided with the true interaction strength and exhibited collective behavior that was absent in the training data. A demonstration with empirical data from a flock of birds showed that AgentNet could identify hidden interaction ranges exhibited by real birds, which cannot be detected by conventional velocity correlation analysis. We expect our framework to open a novel path to investigating complex systems and to provide insight into general process-driven modeling.

arXiv comment17 pages, 8 figures, 3 tables

References in corpus (27)

• Adam: A Method for Stochastic Optimization
• Semi-Supervised Classification with Graph Convolutional Networks
• Novel type of phase transition in a system of self-driven particles
• Epidemic processes in complex networks
• Geometric deep learning: going beyond Euclidean data
• Attention-Based Models for Speech Recognition
• Hidden Physics Models: Machine Learning of Nonlinear Partial Differential Equations
• Scale-free correlations in bird flocks
• Hierarchical group dynamics in pigeon flocks
• Dynamics of Rumor Spreading in Complex Networks
• Data-driven discovery of coordinates and governing equations
• Statistical mechanics for natural flocks of birds
• A Recurrent Latent Variable Model for Sequential Data
• Discovering physical concepts with neural networks
• Dynamic Graph Convolutional Networks
• Learning Multiagent Communication with Backpropagation
• Neural Relational Inference for Interacting Systems
• Generalized energy equipartition in harmonic oscillators driven by active baths
• Flexible Neural Representation for Physics Prediction
• Toward an AI Physicist for Unsupervised Learning
• Cellular automata as convolutional neural networks
• Learning Dynamical Systems from Partial Observations
• Unsupervised Learning of Latent Physical Properties Using Perception-Prediction Networks
• Differentiable Physics-informed Graph Networks
• A comparative study between two models of active cluster-crystals
• Pattern formation with repulsive soft-core interactions: discrete particle dynamics and Dean-Kawasaki equation
• Streaming Graph Neural Networks