Data efficiency of classification strategies for chemical and materials design†

Active learning and design–build–test–learn strategies are increasingly employed to accelerate materials discovery and characterization. Many data-driven materials design campaigns require that materials are synthesizable, stable, soluble, recyclable, or non-toxic. Resources are wasted when materials are recommended that do not satisfy these constraints. Acquiring this knowledge during the design campaign is inefficient, and many materials constraints transcend specific design objectives. However, there is no consensus on the most data-efficient algorithm for classifying whether a material satisfies a constraint. To address this gap, we comprehensively compare the performance of 100 strategies for classifying chemical and materials behavior. Performance is assessed across 31 classification tasks sourced from the literature in chemical and materials science. From these results, we recommend best practices for building data-efficient classifiers, showing the neural network- and random forest-based active learning algorithms are most efficient across tasks. We also show that classification task complexity can be quantified by task metafeatures, most notably the noise-to-signal ratio. These metafeatures are then used to rationalize the data efficiency of different molecular representations and the impact of domain size on task complexity. Overall, this work provides a comprehensive survey of data-efficient classification strategies, identifies attributes of top-performing strategies, and suggests avenues for further study.