A Benchmarking Between Deep Learning, Support Vector Machine and Bayesian Threshold Best Linear Unbiased Prediction for Predicting Ordinal Traits in Plant Breeding

G3: Genes, Genomes, Genetics

article

Osval A. Montesinos-López¹  Javier Martín-Vallejo²  José Crossa³  Daniel Gianola⁴  Carlos M. Hernández-Suárez⁵  Abelardo Montesinos-López⁶  Philomin Juliana³  Ravi Singh³ 

[1] Facultad de Telemática, Universidad de Colima, Colima, Colima, 28040, México;Departamento de Estadística, Universidad de Salamanca, c/Espejo 2, Salamanca, 37007, España;International Maize and Wheat Improvement Center (CIMMYT), Apdo. Postal 6-641, 06600, Ciudad de México, México;Departments of Animal Sciences, Dairy Science, and Biostatistics and Medical Informatics, University of Wisconsin-Madison, Madison,Wisconsin 53706;Facultad de Ciencias, Universidad de Colima, Colima, Colima, 28040, México;Departamento de Matemáticas, Centro Universitario de Ciencias Exactas e Ingenierías (CUCEI),Universidad de Guadalajara, 44430, Guadalajara, Jalisco, México

关键词: threshold;    GBLUP;    deep learning;    support vector machine;    genomic selection;    plant breeding;    Genomic Prediction;    GenPred;    Shared Data Resources;   

DOI  :  10.1534/g3.118.200998

学科分类：社会科学、人文和艺术（综合）

来源: Genetics Society of America

【 摘 要 】

Genomic selection is revolutionizing plant breeding. However, still lacking are better statistical models for ordinal phenotypes to improve the accuracy of the selection of candidate genotypes. For this reason, in this paper we explore the genomic based prediction performance of two popular machine learning methods: the Multi Layer Perceptron (MLP) and support vector machine (SVM) methods vs. the Bayesian threshold genomic best linear unbiased prediction (TGBLUP) model. We used the percentage of cases correctly classified (PCCC) as a metric to measure the prediction performance, and seven real data sets to evaluate the prediction accuracy, and found that the best predictions (in four out of the seven data sets) in terms of PCCC occurred under the TGLBUP model, while the worst occurred under the SVM method. Also, in general we found no statistical differences between using 1, 2 and 3 layers under the MLP models, which means that many times the conventional neuronal network model with only one layer is enough. However, although even that the TGBLUP model was better, we found that the predictions of MLP and SVM were very competitive with the advantage that the SVM was the most efficient in terms of the computational time required.

【 授权许可】

CC BY|CC BY-NC