Akio Onogi

This study uses non-stationary generalized Pareto distributions to estimate changing risks of low crop yield across multiple crops and regions, showing exacerbated trends in several areas and mitigation in Japan for wheat.

This study applies XGBoost with SHAP to QTL mapping, showing comparable performance to traditional methods for main effects and superior performance for interaction effects.

The paper introduces an R package for ensemble learning stacking, enabling one-row scripts for training and prediction.

The paper introduces a Bayesian model that integrates metabolic networks for genomic prediction, showing improved accuracy in simulated and real data.

This simulation study investigates how neural networks predict genetic values associated with gene interactions, finding that pruning redundant markers and using networks with few hidden layers improves accuracy.