Active inference and Bayesian model-based RL learn reward-maximizing behavior in <10 episodes in deterministic FrozenLake.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Active inference recovers performance within 1 episode after context switch in non-stationary FrozenLake, while Bayesian RL requires ~40 episodes.finding0.876
  Figure 4 and discussion in §3.
• Active inference agents engage in information-seeking behavior in reward-free FrozenLake environments, contrasting with Q-learning but similar to Bayesian RL.finding0.875
  Empirical demonstration on FrozenLake; shows epistemic value drives exploration absent reward signal.
• 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.finding0.867
  Key empirical result validating online planning capability of active inference.
• Bayesian model-based RL achieved average score 99.76 [99.45, 100.00] in deterministic FrozenLake.finding0.839
  Table 1.
• Under reward shaping (G=100, H=-100, F=0), Active Inference scored 99.52, Bayesian RL 99.77, Q-learning 95.56, with nearly identical behavior between belief-based agents.finding0.818
  Table 2, row 3, showing equivalence when prior preferences match rewards.
• Active Inference null model (no prior preferences) achieved average score 50.03 [49.70, 50.35] in deterministic FrozenLake.finding0.818
  Table 1.
• Active Inference agent achieved average score 99.88 [99.64, 100.00] in deterministic FrozenLake environment across 200 trials of 500 episodes.finding0.817
  Table 1, deterministic environment row.
• There is an implicit behavioral equivalence between Bayesian model-based reinforcement learning and active inference when prior preferences are treated as a reward function.claim0.800
  §3, reward shaping conclusion.