Adapting Evidential Neural Networks to Test-Time Neighbor Fusion Improves Molecular Property Prediction

arXiv:2607.11091 · cs.LG, q-bio.BM, stat.ML · Submitted 2026-07-13 · Read on arXiv

Cameron Gruich, Weichi Yao, Yixin Wang, Bryan Goldsmith

cs.LG, q-bio.BM, stat.ML

Submitted: 2026-07-13

Comments: 45 pages (18 main, 27 SI); 11 figures (7 main, 4 SI); 14 tables (0 main, 14 SI); 61 equations (15 main, 46 SI)

License: http://creativecommons.org/licenses/by/4.0/

The gist: A trained molecular property model can be refined at test time by correcting each prediction with the measured labels of the most similar training molecules, a retraining-free procedure we call

Terminology

Abstract

A trained molecular property model can be refined at test time by correcting each prediction with the measured labels of the most similar training molecules, a retraining-free procedure we call neighbor fusion; evidential neural networks make it principled by using their aleatoric and epistemic uncertainty to parameterize a Bayesian update. Our main contribution, PG-EVIKAL, learns a property-distance metric to re-rank structurally similar neighbors by their property relevance before fusion, building on EVIKAL (scalar Kalman filter) and GP-EVIKAL (Gaussian process variant handling correlated neighbors). Evaluated on 16 molecular datasets, PG-EVIKAL reduces RMSE relative to the evidential model baseline on 14 of them, with a median reduction of 19.4%, and improves calibration; in sequential-assay scenarios it further incorporates newly measured molecules, refining predictions as they arrive without retraining. This work demonstrates that evidential uncertainty decomposition is not merely a calibration objective but an actionable inference resource that enables test-time refinement of molecular property predictions.

Sources

Related papers