Peano: Learning Formal Mathematical Reasoning (2211.15864v1)
Abstract: General mathematical reasoning is computationally undecidable, but humans routinely solve new problems. Moreover, discoveries developed over centuries are taught to subsequent generations quickly. What structure enables this, and how might that inform automated mathematical reasoning? We posit that central to both puzzles is the structure of procedural abstractions underlying mathematics. We explore this idea in a case study on 5 sections of beginning algebra on the Khan Academy platform. To define a computational foundation, we introduce Peano, a theorem-proving environment where the set of valid actions at any point is finite. We use Peano to formalize introductory algebra problems and axioms, obtaining well-defined search problems. We observe existing reinforcement learning methods for symbolic reasoning to be insufficient to solve harder problems. Adding the ability to induce reusable abstractions ("tactics") from its own solutions allows an agent to make steady progress, solving all problems. Furthermore, these abstractions induce an order to the problems, seen at random during training. The recovered order has significant agreement with the expert-designed Khan Academy curriculum, and second-generation agents trained on the recovered curriculum learn significantly faster. These results illustrate the synergistic role of abstractions and curricula in the cultural transmission of mathematics.
- Thinking fast and slow with deep learning and tree search. Advances in Neural Information Processing Systems, 30, 2017.
- HOList: An environment for machine learning of higher order logic theorem proving. In Kamalika Chaudhuri and Ruslan Salakhutdinov, editors, Proceedings of the 36th International Conference on Machine Learning, volume 97 of Proceedings of Machine Learning Research, pages 454–463. PMLR, 09–15 Jun 2019.
- Curriculum learning. In Proceedings of the 26th annual international conference on machine learning, pages 41–48, 2009.
- On the properties of neural machine translation: Encoder–decoder approaches. Syntax, Semantics and Structure in Statistical Translation, page 103, 2014.
- On the notion of interestingness in automated mathematical discovery. International Journal of Human-Computer Studies, 53(3):351–375, 2000.
- A machine program for theorem-proving. Communications of the ACM, 5(7):394–397, 1962.
- A computing procedure for quantification theory. Journal of the ACM (JACM), 7(3):201–215, 1960.
- Dreamcoder: Bootstrapping inductive program synthesis with wake-sleep library learning. In Proceedings of the 42nd acm sigplan international conference on programming language design and implementation, pages 835–850, 2021.
- William Timothy Gowers. The two cultures of mathematics. Mathematics: frontiers and perspectives, 65:65, 2000.
- A formal proof of the kepler conjecture. In Forum of mathematics, Pi, volume 5. Cambridge University Press, 2017.
- Jesse Michael Han and Floris van Doorn. A formal proof of the independence of the continuum hypothesis. In Proceedings of the 9th ACM SIGPLAN International Conference on Certified Programs and Proofs, pages 353–366, 2020.
- Gamepad: A learning environment for theorem proving. In International Conference on Learning Representations, 2018.
- Deepmath-deep sequence models for premise selection. Advances in neural information processing systems, 29, 2016.
- Reinforcement learning of theorem proving. Advances in Neural Information Processing Systems, 31, 2018.
- Maurice G Kendall. Rank correlation methods. 1955.
- Imagenet classification with deep convolutional neural networks. Communications of the ACM, 60(6):84–90, 2017.
- Intuitionistic type theory, volume 9. Bibliopolis Naples, 1984.
- A proposal for the dartmouth summer research project on artificial intelligence, august 31, 1955. AI magazine, 27(4):12–12, 2006.
- Machines who think: A personal inquiry into the history and prospects of artificial intelligence. CRC Press, 2004.
- Playing atari with deep reinforcement learning. arXiv preprint arXiv:1312.5602, 2013.
- Report on a general problem solving program. In IFIP congress, volume 256, page 64. Pittsburgh, PA, 1959.
- Frank Pfenning. Unification and anti-unification in the calculus of constructions. In LICS, volume 91, pages 74–85, 1991.
- Frank Pfenning. Elf: A meta-language for deductive systems. In International Conference on Automated Deduction, pages 811–815. Springer, 1994.
- Gordon Plotkin. Automatic methods of inductive inference. 1972.
- Contrastive reinforcement learning of symbolic reasoning domains. Advances in Neural Information Processing Systems, 34:15946–15956, 2021.
- Formal mathematics statement curriculum learning. arXiv preprint arXiv:2202.01344, 2022.
- John Alan Robinson. Theorem-proving on the computer. Journal of the ACM (JACM), 10(2):163–174, 1963.
- Stuart J Russell. Artificial intelligence a modern approach. Pearson Education, Inc., 2010.
- A general reinforcement learning algorithm that masters chess, shogi, and go through self-play. Science, 362(6419):1140–1144, 2018.
- Michael Tomasello. The cultural origins of human cognition. Harvard university press, 2009.
- Daniel Whalen. Holophrasm: a neural automated theorem prover for higher-order logic. arXiv preprint arXiv:1608.02644, 2016.
- Tacticzero: Learning to prove theorems from scratch with deep reinforcement learning. Advances in Neural Information Processing Systems, 34:9330–9342, 2021.