Applications of Machine Learning in Breeding for Stress Tolerance in Maize.

ORNELLA LEONARDO; CERVIGNI GERARDO; TAPIA ELIZABETH

Crop Stress and its Management: Perspectives and Strategies

Springer

Año: 2011; p. 163 - 192

Corn is one of the world?s most important cereals and a major source of calories for humanity, along with rice and wheat. Climate change and the use of marginal land for crop production require the development of genotypes adapted to stressful environments, particularly drought tolerant plants. Among the new technologies currently available for accelerate the releasing of new genotypes there is an emerging discipline called Machine Learning (ML). A primary goal of ML algorithms is to automatically learn to recognize complex patterns and make intelligent decisions based on data. This work reviews several strategic applications of ML in maize breeding. Quantitative trait loci mapping, heterotic group assignment and the popular genome-wide selection are some of the key areas currently addressed by the literature. Results are encouraging and propose ML algorithms as a valuable alternative to traditional statistical techniques applied in maize, even the more recently introduced linear mixed models.