Neurosymbolic programming for science
A 2022 position paper, presented at the NeurIPS 2022 AI for Science workshop, on combining neural learning with symbolic programs. It aims to give researchers interpretable machine-learning tools.
Consulting service
Neurosymbolic AI Technical Advisory
Technical guidance for projects combining statistical learning with symbolic programs, reusable abstractions, or equation search. Scope can include system design, interpretable outputs, and evaluation for a defined task.
Scope
What this advisory can cover
Plan program-synthesis and neural-symbolic system designs
Represent domain knowledge, constraints, and reusable concepts
Develop or review domain-specific languages for a defined task
Design search and evaluation strategies for generated programs or equations
Make candidate models and reasoning steps easier to inspect
Assess where a method is suitable for scientific model discovery
Related work
Published work related to this advisory area
Selected publications, talks, and projects related to this area. Each link opens the work and its original source.
Lemma: Bootstrapping high-level mathematical reasoning with learned symbolic abstractions
LEMMA, a 2022 arXiv paper, teaches reinforcement-learning agents reusable, higher-level symbolic actions for solving equations and simplifying fractions.
MathDSL: A Domain-Specific Language for Concise Mathematical Solutions Via Program Synthesis
MathDSL is a domain-specific language that lets the DreamCoder program-synthesis system write short, readable equation-solving programs. It is a 2024 arXiv paper led by Sagnik Anupam.
InceptionSR: Recursive Symbolic Regression for Equation Synthesis
InceptionSR is a symbolic-regression method that freezes promising partial equations and reuses them as features in later search rounds. It is tested on planetary-system instability and a triangle-area formula.