Simplicity Bias in Transformers and their Ability to Learn Sparse Boolean Functions.

Satwik Bhattamishra Arkil Patel Varun Kanade Phil Blunsom

Published in: ACL (1) (2023)

Keyphrases

boolean functions
uniform distribution
prime implicants
dnf formulae
threshold functions
high dimensional
membership queries
multi valued
functional properties
bi decomposition
dnf formulas
linear threshold
pseudo boolean functions
truth table
disjunctive normal form
efficient learning
genetic programming