Agent-based ecosystem modeling with deep reinforcement learning

doi:10.1016/j.ecoinf.2026.103819

Agent-based ecosystem modeling with deep reinforcement learning

2026-05-15

0

OA

AI

DOI：10.1016/j.ecoinf.2026.103819

Original

Save

View PDF

Abstract

En 中文

• Deep reinforcement learning replaced hand-coded rules in an agent-based ecosystem model. • Agents were trained across environments using a homeostatic reward. • The framework was applied to an Alpine wolf-chamois-vegetation system. • Simulations reproduced coexistence, predator–prey cycles, and habitat use. • The model remained resilient to moderate hunting and habitat degradation.

Keywords:

Agent-based modeling

Deep reinforcement learning

Ecosystem modeling

Pattern-oriented modeling

Sustainable decision-making

Agent-based ecosystem modeling with deep reinforcement learning

Abstract

AI Summary

Journal

Organization