Prediction of bitterant and sweetener using structure-taste relationship models based on an artificial neural network

Identifying the taste characteristics of molecules is essential for their application in foods and drugs. We developed three structure-taste relationship models using convolutional neural networks (CNN), multi-layer perceptron with descriptors (MLP-Descriptor), and MLP with fingerprints (MLP-Fingerprint) to classify compounds. The MLP-Fingerprint model showed the highest predictive accuracy. The MLP-Descriptor model provided a clear structure-taste relationship, while the CNN model offered efficient feature extraction from 2D chemical maps. These computational methods provide a valuable technique for identifying molecular taste.

Recommended citation: Bo, W., Qin, D., Zheng, X., Wang, Y., Ding, B., Li, Y., & Liang, G. (2022). "Prediction of bitterant and sweetener using structure-taste relationship models based on an artificial neural network." Food Research International. 153:110974.
Download Paper