Holophrasm: a neural Automated Theorem Prover for higher-order logic

I propose a system for Automated Theorem Proving in higher order logic using deep learning and eschewing hand-constructed features. Holophrasm exploits the formalism of the Metamath language and explores partial proof trees using a neural-network-augmented bandit algorithm and a sequence-to-sequence model for action enumeration. The system proves 14% of its test theorems from Metamath's set.mm module.

[1]  Norman D. Megill,et al.  Metamath A Computer Language for Pure Mathematics , 1969 .

[2]  John Harrison,et al.  HOL Light: A Tutorial Introduction , 1996, FMCAD.

[3]  Christine Paulin-Mohring,et al.  The coq proof assistant reference manual , 2000 .

[4]  Thomas C. Hales,et al.  Introduction to the Flyspeck Project , 2005, Mathematics, Algorithms, Proofs.

[5]  Stephan Merz,et al.  Proving the Correctness of Disk Paxos , 2005, Arch. Formal Proofs.

[6]  R. Matuszewski,et al.  M IZAR : the first 30 years , 2005 .

[7]  Csaba Szepesvári,et al.  Bandit Based Monte-Carlo Planning , 2006, ECML.

[8]  Yoshua Bengio,et al.  Understanding the difficulty of training deep feedforward neural networks , 2010, AISTATS.

[9]  Josef Urban,et al.  Overview and Evaluation of Premise Selection Techniques for Large Theory Mathematics , 2012, IJCAR.

[10]  Cezary Kaliszyk,et al.  MaSh: Machine Learning for Sledgehammer , 2013, ITP.

[11]  Jesse Alama,et al.  Premise Selection for Mathematics by Corpus Analysis and Kernel Methods , 2011, Journal of Automated Reasoning.

[12]  Quoc V. Le,et al.  Sequence to Sequence Learning with Neural Networks , 2014, NIPS.

[13]  David Silver,et al.  Move Evaluation in Go Using Deep Convolutional Neural Networks , 2014, ICLR.

[14]  Cezary Kaliszyk,et al.  MizAR 40 for Mizar 40 , 2013, Journal of Automated Reasoning.

[15]  Christopher D. Manning,et al.  Effective Approaches to Attention-based Neural Machine Translation , 2015, EMNLP.

[16]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[17]  Demis Hassabis,et al.  Mastering the game of Go with deep neural networks and tree search , 2016, Nature.

[18]  Josef Urban,et al.  DeepMath - Deep Sequence Models for Premise Selection , 2016, NIPS.