Optimal Allocation of Embedding Dimensions under Finite-Sample Constraints

Vasileios E. Papageorgiou

The embedding dimension of categorical predictors is usually selected through heuristic tuning, although it directly affects model complexity, approximation quality, and finite-sam…