Benchmarking hyperparameter optimization strategies for crop genomic prediction.

Bayesian optimization consistently achieves high prediction accuracy and stability across multiple crop datasets, especially in rice and wheat, outperforming other strategies in genomic prediction.

• Why it matters: Optimizing hyperparameters is crucial for improving genomic prediction models, yet its impact has been underexplored in crop breeding, limiting the efficiency of predictive strategies.
• What they did: The study evaluated six genomic prediction models across four crop datasets (maize, rice, wheat, foxtail millet) using four hyperparameter optimization strategies, assessing prediction performance and computational efficiency.
• The result: Findings suggest that the best optimization strategy depends on crop species, traits, and models, with Bayesian optimization offering a balance of accuracy and stability, guiding more effective crop breeding decisions.