Skip to main content

Table 15 Description of the neural networks’ parameters used for the experiment

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

         

RMSE Threshold

Neural Net

Number of Layers Node

Number of Input Nodes

Number of Hidden Nodes

Number of Output

η

α

Weight Values Range

Epoch

Convergence

Training

ANN

3

Same as feature vector

2

1

0.9

0.3

(−0.3,0.3)

50,000

0.02

0.05

NGN

Dynamic

Dependant on input token

12

1

0.3

0.3

(−1.6,−1.0) or (1.0,1.6)

10,000

0.03

0.05

  1. *activation value of 1.0
  2. *hidden nodes were tested for best performance, leading to 12 hidden nodes to be used with the NGN for the toxicity dataset