With our proof of concept we came to the following conclusion: The correlation (3/3) is the proof of concept: we can predict the smell of a component from its chemical composition. This is a major step in product design and product development because it helps to assess which component has what effect in the creation of a new product!
Several scientific papers relating to odor perception are listed here below and are linked to our first PoC software:
Some are linked to the Nakamoto Laboratory witch which we have had an indrect collaboration with at the Tokyo Institute of Technology:
Predicting human olfactory perception from chemical features of odor molecules
https://www.science.org/doi/10.1126/science.aal2014
Predicting Odor Perception of Mixed Scent from Mass Spectrometry https://iopscience.iop.org/article/10.1149/1945-7111/ac33e0
Predictive modeling for odor character of a chemical using ML combined with NLP
Predicting human odor perception represented by continuous values from mass spectra of essential oils https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0234688
Learning Generalizable Perceptual Representations of Small Molecules https://arxiv.org/abs/1910.10685
Transport features predict if a molecule is odorous https://www.pnas.org/doi/10.1073/pnas.2116576119
Predicting individual perceptual scent impression from imbalanced dataset using mass spectrum of odorant molecules https://www.nature.com/articles/s41598-022-07802-3
Thank you for your attention! Enjoy the reading!
The Ajinomatrix Team