Speaker: Dr. Cezary Kaliszyk (University of Innsbruck, Austria)
Title: Automating Formalization using Statistical and Neural Approaches
Abstract:
The talk will give a summary of our experiments to automatically
translate informal mathematical documents to proof formalizations.
We will first introduce a context-based parsing approach that
combined statistical deep parse trees with type checking and
large-theory automated reasoning. We will next discuss the use
of a neural translation model that automatically translates
informalized Latex-written text to formal proof assistant languages.
Parts of the work have been submitted as a joint paper with Qingxiang
Wang and Josef Urban to CICM'18.
Back