{"slug": "from-black-box-to-executable-logic-explainable-reinforcement-learning-through", "title": "From Black Box to Executable Logic: Explainable Reinforcement Learning through Prolog Expert Systems", "summary": "Researchers from an undisclosed institution have developed a method to transform a trained deep reinforcement learning policy into an executable Prolog logic program that is readable, runnable, and editable, achieving exact optimal return on a two-room key-and-door task with 16,944 reachable states and matching the neural teacher within noise on Acrobot with eleven clauses. The three-stage post-hoc transformation extracts a frozen proximal policy optimization teacher, induces an ordered rule list, and emits a Prolog program with guarantees including a return-loss bound and monotonic improvement. The approach recovers about 97% of the neural teacher's return on CartPole but only partially on LunarLander, consistent with an exponential lower bound in observation dimension.", "body_md": "arXiv:2607.15459v1 Announce Type: new\nAbstract: A trained deep reinforcement learning policy is a black box, and we ask whether it can be made explainable by rewriting it as an executable logic program that reproduces its behaviour and that a person can read, a logic engine can run, and an optimizer can edit. We present a three-stage post-hoc transformation that extracts a frozen proximal policy optimization teacher, induces an ordered rule list from its decisions in the manner of classical relational learning, and emits the result as a Prolog program whose every decision is executed by an off-the-shelf logic engine; a subsequent expansion stage edits the rule base and accepts an edit only when policy evaluation certifies a return increase. We prove four guarantees. A return-loss bound makes the distilled program a machine-checkable certificate in a finite Markov decision process, and the expansion loop improves monotonically and terminates. For the continuous-observation setting we answer whether the conversion is possible at all: the propositional threshold instantiation converts the network to arbitrary fidelity as the resolution B grows, with disagreement O(1/B) and a return gap that closes at the same rate, and a matching lower bound shows the cost is exponential in the observation dimension for an oblique decision boundary. Empirically, on a two-room key-and-door task with 16,944 reachable states the expanded Prolog program attains exact optimal return in every seed and, in a budget-capped regime, exceeds the stochastic teacher on exact return in ten of ten seeds. On three continuous-control tasks the emitted program substitutes the network, matching the neural teacher within noise on Acrobot with eleven clauses and recovering about 97% of its return on CartPole, while on the finer-control LunarLander it recovers only partially, exactly the ceiling the exponential lower bound predicts.", "url": "https://wpnews.pro/news/from-black-box-to-executable-logic-explainable-reinforcement-learning-through", "canonical_source": "https://arxiv.org/abs/2607.15459", "published_at": "2026-07-20 04:00:00+00:00", "updated_at": "2026-07-20 13:52:55.963669+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "ai-research"], "entities": ["arXiv", "Prolog"], "alternates": {"html": "https://wpnews.pro/news/from-black-box-to-executable-logic-explainable-reinforcement-learning-through", "markdown": "https://wpnews.pro/news/from-black-box-to-executable-logic-explainable-reinforcement-learning-through.md", "text": "https://wpnews.pro/news/from-black-box-to-executable-logic-explainable-reinforcement-learning-through.txt", "jsonld": "https://wpnews.pro/news/from-black-box-to-executable-logic-explainable-reinforcement-learning-through.jsonld"}}