Conceptlearned density model / 学习名义分布
Learned Density Model
A probability density model learned from data that assigns positive probability density to unobserved realizations, thereby supporting structural extrapolation. In KL-DRO it is used as the nominal distribution, allowing the optimization to retain the ability to generalize beyond the empirical samples rather than hedging only against observed data.
Timeline (last 90 days)
Oct 30
When the KL ball is centered at the empirical distribution, distributions within the ball can only reweight historical records; when it is centered at a learned density model, the model may assign positive density to unobserved realizations, thereby allowing the adversary to increase the weight of these realizations.
Unverified50%