Co-Diffusion: An Affinity-Aware Two-Stage Latent Diffusion Framework for Generalizable Drug-Target Affinity Prediction

A set of new ML/statistics papers covering theoretical and applied advances: Co-Diffusion introduces an affinity-aware two-stage latent diffusion framework to improve drug–target affinity (DTA) prediction and zero-shot generalization across unseen molecules and proteins; a height-augmented ReLU network design gives more parameter-efficient exponential approximation rates for analytic and L^p functions; deep regression with minimum error entropy (MEE) yields minimax-optimal rates under strong mixing; diagnostic transport maps provide covariate-dependent recalibration and local diagnostics for (