Feldman
A machine learning theory researcher who proposed a lower-bound result on the cost of memorization, attempting to characterize the relationship between sample frequency and the memorization penalty with a quantitative coefficient.
Timeline (last 90 days)
A recent corrective study gives an exact three-point counterexample, proving that the frequency coefficients in Feldman's memorization lower bound fail in both the conditional form and the averaged form.
The frequency coefficient used by Feldman is actually the posterior-mean form of Good's exact frequency formula, and can be written as the occupancy-weighted average of the corrected coefficient.
The correction does not negate Feldman's work, but places it within a more precise framework; the original coefficient is an approximation of the corrected coefficient under a specific weighting.
All Facts (3)
A recent corrective study gives an exact three-point counterexample, proving that the frequency coefficients in Feldman's memorization lower bound fail in both the conditional form and the averaged form.
50%UnverifiedThe frequency coefficient used by Feldman is actually the posterior-mean form of Good's exact frequency formula, and can be written as the occupancy-weighted average of the corrected coefficient.
50%UnverifiedThe correction does not negate Feldman's work, but places it within a more precise framework; the original coefficient is an approximation of the corrected coefficient under a specific weighting.
50%