CORTEXA
← Browse

Lucas Sheneman

1 paper indexed

arxivcs.LGcs.AIcs.PL2026-07-03

Differentiate the Evaluator, Not the Program: An Efficient Runtime Representation for Neuro-Symbolic Learning

Lucas Sheneman

AI systems increasingly propose executable scientific models whose value depends on both their symbolic structure and their fitted continuous parameters. This makes parameter calibration the bottleneck of program-and-parameter co-search: an outer loop can generate thousands of ca…

View free PDFSource page