New VitisGP dataset benchmarks machine learning for grapevine trait prediction

Researchers have introduced VitisGP, a new multi-phenotype dataset designed to predict grapevine characteristics from genomic data. This dataset integrates genome-wide SNP data with 26 different prediction tasks, including aroma, botanical, and wine color traits, across 1,197 Vitis vinifera cultivars. The study benchmarks various machine learning models, finding that CatBoost performed best overall with a macro-average AUROC of 0.696 across all tasks, though performance varied by trait type. AI

IMPACT This dataset and benchmark could advance the application of machine learning in agriculture for crop breeding and climate adaptation.

RANK_REASON The item describes a new dataset and benchmark for a specific biological prediction task using machine learning models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• CatBoost
• GBLUP
• L2-regularized logistic regression
• random forest
• Transformer-based Models
• VitisGP
• Vitis vinifera
• XGBoost

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →