Machine Learning for Scent: Learning Generalizable Perceptual Representations of Small Molecules
This addresses the quantitative structure-odor relationship problem for applications in chemistry, fragrance, and neuroscience, representing an incremental advance with domain-specific impact.
The paper tackled predicting molecular odor from structure using graph neural networks, achieving significant performance improvements over prior methods on a novel expert-labeled dataset and demonstrating meaningful odor space representations through strong transfer learning results.
Predicting the relationship between a molecule's structure and its odor remains a difficult, decades-old task. This problem, termed quantitative structure-odor relationship (QSOR) modeling, is an important challenge in chemistry, impacting human nutrition, manufacture of synthetic fragrance, the environment, and sensory neuroscience. We propose the use of graph neural networks for QSOR, and show they significantly out-perform prior methods on a novel data set labeled by olfactory experts. Additional analysis shows that the learned embeddings from graph neural networks capture a meaningful odor space representation of the underlying relationship between structure and odor, as demonstrated by strong performance on two challenging transfer learning tasks. Machine learning has already had a large impact on the senses of sight and sound. Based on these early results with graph neural networks for molecular properties, we hope machine learning can eventually do for olfaction what it has already done for vision and hearing.