English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The training of molecular models of quantum mechanical properties based onstatistical machine learning requires large datasets which exemplify the mapfrom chemical structure to molecular property. Intelligent a priori selectionof training examples is often difficult or impossible to achieve as priorknowledge may be sparse or unavailable. Ordinarily representative selection oftraining molecules from such datasets is achieved through random sampling. Weuse genetic algorithms for the optimization of training set compositionconsisting of tens of thousands of small organic molecules. The resultingmachine learning models are considerably more accurate with respect to smallrandomly selected training sets: mean absolute errors for out-of-samplepredictions are reduced to ~25% for enthalpies, free energies, and zero-pointvibrational energy, to ~50% for heat-capacity, electron-spread, andpolarizability, and by more than ~20% for electronic properties such asfrontier orbital eigenvalues or dipole-moments. We discuss and presentoptimized training sets consisting of 10 molecular classes for all molecularproperties studied. We show that these classes can be used to design improvedtraining sets for the generation of machine learning models of the sameproperties in similar but unrelated molecular sets.

Genetic optimization of training sets for improved machine learning models of molecular properties