Alfa-Pi-Mi Intelligent Neural Network Model: optimizing learning in engineering through artificial intelligence

Abstract

This paper presents the design and implementation of the Alfa-Pi-Mi neural network model, developed to personalize and optimize learning for Industrial Engineering students through artificial intelligence (AI). The model identifies learning patterns and generates tailored instructional strategies. The findings should be interpreted as initial feasibility evidence from a pilot study, not as generalizable effects. We detail the system architecture (input variables, activation functions, and optimization methods) and report pilot results with n = 30 students over four weeks: academic performance improved relative to baseline in Optimization and Mathematical Models (+60%) and Statistics and Quantitative Methods (+55%), with additional gains in Production Management (+50%), Project Management (+45%), and Quality Engineering (+50%). Overall, 62% of participants achieved moderate-to-high improvement, and learning trajectories steepened from week 3. In addition, as a methodological validation of the predictive component, a minimal multilayer perceptron (MLP) task reduced mean absolute error (MAE) by 23% and root mean squared error (RMSE) by 19% versus baseline (initial, untuned configuration), after normalization/standardization and hyperparameter tuning.