A Reverse-Gold Theorem: Positive-Data Identification by Substrate-Bounded Contrastive Learners
stat.ML, cs.CL, cs.LG, q-bio.NC
Submitted: 2026-10-20
Updated: 2026-10-20
Related papers
- Behavior of prediction performance metrics with rare events
- Optimal Estimation of Generic Dynamics by Path-Dependent Neural Jump ODEs
- A Posterior-Dynamics Framework for Imaging Inverse Problems with Pretrained Diffusion Priors
- Efficient quadratic entropy with distance sketches
- Embedding-Bias in Conditional Independence Testing
- From Log-Odds to Shapley Values: An Explanatory Geometry for the Weighted Naive Bayes Classifier