Research Article

Application of Artificial Intelligence to the Prediction of the Antimicrobial Activity of Essential Oils

Table 4

Effect of different learning sets on the prediction by an artificial neural network (FANN) of the inhibition diameter in disk diffusion assays for S. aureus (1 output, small data input set). Accuracy is expressed in the percentage of predictions within the given error margin.

Learning setAccuracy (%)
ΔID ≤ 5 mmΔID ≤ 10 mmΔID ≤ 15 mmΔID > 15 mm

Higher %50.066.783.316.7
Lower %37.566.791.78.3
Random45.879.291.78.3