Cheminformatics-based machine learning (ML) has assisted in determining optimal reaction conditions, including catalyst struc-tures, in the field of synthetic chemistry. However, such ML-focused strategies have remained largely unexplored in the context of catalytic molecular transformations using Lewis-acidic main-group elements, probably due to the absence of a candidate library and effective guidelines (parameters) for the prediction of the activity of main-group elements. Here, the construction of a triarylborane library and its application to an ML-assisted approach for the catalytic reductive alkylation of amino acids with aldehydes and H2 is reported. The obtained results suggest that the deformation energy serves as a useful parameter for Gauss-ian process regression to construct adequate models for predicting the turnover frequencies of triarylboranes under the applied model reaction conditions, while Gaussian progress regression based on the energy levels of the lowest unoccupied molecular orbitals (LUMOs) including the empty p-orbitals of boron tend to underestimate the turnover frequency. The optimal borane, i.e., B(2,3,5,6-Cl4-C6H)(2,6-F2-3,5-(CF3)2-C6H)2, effectively catalyzes the reductive functionalization of aniline-derived amino acids and C-terminal-protected peptides in the presence of 4-methyltetrahydropyrane and H2, generating H2O as the sole by-product.