Report
How FEGF and DAM reportedly differ on geodesic preprocessing
In a reply, a poster says FEGF uses JKO conditions and geodesic preprocessing, while DAM avoids that preprocessing.
TLDR
A poster says Rancati, Maas and Francesco Locatelli published a method they call FEGF (ICML 2026). They describe it as bridging the math with machine learning through JKO conditions and geodesic preprocessing. The poster claims DAM avoids that preprocessing entirely.
Combined views
—
2 Sources, first seen ago
