Q-learning (epsilon=1 decaying to 0) achieved average score 80.44 [78.96, 81.93] in deterministic FrozenLake.

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

• In the absence of any reward signal, Q-learning (epsilon=0.1) learns a deterministic circular policy with score 0.00 and does not explore purposefully.finding0.840
  Table 2 first row; reward shaping section.
• Bayesian model-based RL achieved average score 99.76 [99.45, 100.00] in deterministic FrozenLake.finding0.817
  Table 1.
• Active Inference null model (no prior preferences) achieved average score 50.03 [49.70, 50.35] in deterministic FrozenLake.finding0.800
  Table 1.
• Active Inference agent achieved average score 99.88 [99.64, 100.00] in deterministic FrozenLake environment across 200 trials of 500 episodes.finding0.791
  Table 1, deterministic environment row.
• Active inference and Bayesian model-based RL learn reward-maximizing behavior in <10 episodes in deterministic FrozenLake.finding0.781
  Discussion of Figure 3.
• In the absence of prior preferences, Active Inference null model and Bayesian RL maintain exploration with average scores of 44.00 and 39.94 respectively, whereas Q-learning does not explore.finding0.766
  Table 2 first row; reward shaping section.
• 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.finding0.765
  Table 2, row 3, showing equivalence when prior preferences match rewards.
• 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.finding0.745
  Key empirical result validating online planning capability of active inference.