Skip to main content

Table 4 Ablation study of individual components of NLLCAL

From: A hybrid framework for improving uncertainty quantification in deep learning-based QSAR regression modeling

Metrics

Combinations

Splitting strategies

IVIT

IVOT

OVOT

SCC

L

0.281

0.202

0.174

M + E

0.273

0.197

0.177

M + E + L + F

0.308

0.225

0.194

ECE

L

0.035

0.065

0.062

M + E

0.036

0.074

0.059

M + E + L + F

0.030

0.058

0.059

ENCE

L

0.192

0.337

0.272

M + E

0.188

0.377

0.249

M + E + L + F

0.184

0.287

0.258

  1. Each letter represents an individual method: M refers to MVE, E refers to ENS, L refers to LDIST and F refers to FDIST
  2. Best results are highlighted in boldface type for each splitting strategy
Back to article page

Journal of Cheminformatics

ISSN: 1758-2946

Contact us