Machine Learning May Sometimes Simply Capture LiteraturePopularity Trends: A Case Study of Heterocyclic Suzuki-MiyauraCoupling

Applications of machine learning (ML) to synthetic chemistry rely on the assumption that large numbers ofliterature-reported examples should enable construction of accurate and predictive models of chemical reactivity. This paperdemonstrates that abundance of carefully curated literature data may be insufficient for this purpose. Using an example of Suzuki-Miyaura coupling with heterocyclic building blocks & xe0d5;and a carefully selected database of >10,000 literature examples & xe0d5;we show thatML models cannot offer any meaningful predictions of optimum reaction conditions, even if the search space is restricted to onlysolvents and bases. This result holds irrespective of the ML model applied (from simple feed-forward to state-of-the-art graph-convolution neural networks) or the representation to describe the reaction partners (variousfingerprints, chemical descriptors,latent representations, etc.). In all cases, the ML methods fail to perform significantly better than naive assignments based on thesheer frequency of certain reaction conditions reported in the literature. These unsatisfactory results likely reflect subjectivepreferences of various chemists to use certain protocols, other biasing factors as mundane as availability of certain solvents/reagents,and/or a lack of negative data. Thesefindings highlight the likely importance of systematically generating reliable and standardizeddata sets for algorithm training.