All three agent types (active inference, Q-learning, Bayesian RL) perform adequately in stationary FrozenLake; only active inference achieves Bayes-optimal behavior in non-stationary settings.

{
  "doc": "sajid_2021_active_inference_demystified.md",
  "author": null,
  "context": "Key empirical result validating online planning capability of active inference.",
  "enrichment": {
    "is_stale": true
  },
  "norm_label": "All three agent types (active inference, Q-learning, Bayesian RL) perform adequately in stationary FrozenLake; only active inference achieves Bayes-optimal behavior in non-stationary settings.",
  "source_url": null,
  "graphify_id": "finding_nonstationarity",
  "source_file": "sajid_2021_active_inference_demystified.md",
  "imported_from": "/Users/antonborzov/Documents/Research.nosync/papers/extract_typed_out/sajid_2021_active_inference_demystified/graph.json",
  "extracted_type": "finding",
  "source_location": "§1 Introduction, L232-234",
  "graphify_file_type": "finding"
}