An innovative approach that amalgamates the strengths of Full Information Maximum Likelihood (FIML) estimation with the capabilities of Self-Attention neural networks.

Our comprehensive experiments on both simulated and real-world datasets underscore SESA’s pronounced advantages over traditional baseline techniques, encapsulating facets of accuracy, computational efficiency, and adaptability to diverse data structures. Especially on small and middle size dataset.

In the folder /paper, or see it at arXiv.