PlayingWORDLE with UPPAAL STRATEGO
| Autor: | Peter G. Jensen, Kim G. Larsen, Marius Mikučionis |
|---|---|
| Přispěvatelé: | Jansen, Nils, Stoelinga, Mariëlle, van den Bos, Petra |
| Jazyk: | angličtina |
| Rok vydání: | 2022 |
| Zdroj: | Jensen, P G, Larsen, K G & Mikučionis, M 2022, PlayingWORDLE with UPPAAL STRATEGO . in N Jansen, M Stoelinga & P van den Bos (eds), A Journey from Process Algebra via Timed Automata to Model Learning : Essays Dedicated to Frits Vaandrager on the Occasion of His 60th Birthday . Springer, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13560 LNCS, pp. 283-305 . https://doi.org/10.1007/978-3-031-15629-8_15 Lecture Notes in Computer Science ISBN: 9783031156281 |
| DOI: | 10.1007/978-3-031-15629-8_15 |
| Popis: | In this paper we model and solve the popular game Wordle using Uppaal Stratego. We model three different game-modes in terms of POMDPs, with more than 12,000 controllable actions. These constitute by far the largest models ever presented to Uppaal Stratego. Our experimental evaluation is encouraging: e.g. in the hard game-mode the partitioning-refinement learning method of Uppaal Stratego reduces the expected number of guesses from a baseline of 7.67 to 4.40 using 1 million training episodes. To better understand the convergence properties of our learning method we also study reduced versions of Wordle. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|