Skip to main content

Table 9 A comparison of the work done by [3] and [4] against the InChI-NGN for the FXa dataset. Missing values are due to lacking information in the primary literature.

From: Classifying and scoring of molecules with the NGN: new datasets, significance tests, and generalization

Design

Method

Q(%)

SE(%)

SP(%)

MCC

Train, Test

A-MIF

88

-

-

-

 

MIF-MIF

84

-

-

-

 

MK1

94.5

98.7

89.5

-

 

MK2

95.2

98.9

87.7

-

 

InChI-NGN

83.8

84.3

82.9

0.657

Leave-20%-Out

InChI-NGN

86.4±2.50

88.5±0.87

82.7±0.05

0.705±0.052