Authors derive a full KL-error decomposition via refined information geometry on mode interactions, then introduce the MAHGenTa algorithm that performs sparse greedy selection of those interactions for improved log-likelihood on limited data.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
A Complete Decomposition of KL Error using Refined Information and Mode Interaction Selection
Authors derive a full KL-error decomposition via refined information geometry on mode interactions, then introduce the MAHGenTa algorithm that performs sparse greedy selection of those interactions for improved log-likelihood on limited data.