Optimization of individual educational trajectories based on hybrid neural network architectures in the context of digital transformation

The paper theoretically substantiates the mechanisms of parametric optimization for personalized educational trajectories using hybrid models based on neural networks (NN). In the context of intensive digital transformation within the educational sector, the transition to adaptive learning management systems is investigated, where the integration of recurrent (RNN), graph (GNN), and transformer (Transformers) architectures provides high accuracy in modeling cognitive progress. The functional role of dynamic content adjustment and predictive analysis of mental patterns as fundamental components of the optimization strategy is revealed. A number of technical and epistemological limitations are identified, in particular, the problem of algorithmic opacity, necessitating the implementation of hybrid models to enhance the interpretability of predictive conclusions. Particular attention is paid to the strategic importance of optimized artificial intelligence (AI) systems in Ukraine as a means of mitigating learning gaps during crisis periods. Table: 1. Ref.: 22 items.