0642813 20260213094651.320251209235959.9 105018304037 10.1007/978-3-032-05134-9_8 14 s. P cav_un_epca*0639225978-3-032-05133-20302-9743104-117ChamSpringer2026Sauerwald K.ThimmM. Restricted Boltzmann Machines BN2A models Educational Assessment Interpretable machine learning cav_un_auth*0458433 Pérez Cabrera Iván UTIA-B Zpracování signálů Department of Signal Processing ZS ZS MX Ústav teorie informace a automatizace AV ČR, v. v. i. cav_un_auth*0101228 Vomlel Jiří UTIA-B Matematická teorie rozhodování Department of Decision Making Theory MTR MTR Department of Decision Making Theory Ústav teorie informace a automatizace AV ČR, v. v. i. cav_un_auth*0226898 Martinková P. CZ https://library.utia.cas.cz/separaty/2025/ZS/perez-0642813.pdf GA25-18070S GA ČR CZ cav_un_auth*0484465 Restricted Boltzmann Machines (RBMs) are bipartite graphical models with binary latent and observed variables that have shown promise for representation learning. However, their lack of interpretable parameters limits their utility in domains requiring explainability, like educational assessment. Despite extensive RBM research, non-negativity constraints on weights—essential for monotonicity in educational contexts—remain largely unexplored. To address this, we propose a method to translate RBMs into a specialized class of bipartite Bayesian networks, which we term BN2A networks, characterized by strict 2-layer separation (hidden and observed variables), Noisy-AND conditional probability tables, and directly interpretable parameters for educational models. Our work establishes a mathematical transformation from RBM weights to BN2A’s interpretable parameters (leak and penalty probabilities), theoretical analysis showing BN2A’s constrained connectivity is a subset of RBM architectures, and empirical evidence that the transformation preserves model fidelity under realistic conditions. By bridging these paradigms, our method leverages RBM’s representational power while achieving BN2A’s interpretability, opening new possibilities for adaptive learning systems and diagnostic tools. cav_un_auth*0498692 European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty, ECSQARU 2025 /19./ 20250923 20250926 Hagen DE 2027 IN 10000 10200 10201 PR RVO:67985556 https://hdl.handle.net/11104/0372668 S 2026 105018304037 SCOPUS PUBMED WOS cav_un_epca*0639225 Symbolic and Quantitative Approaches to Reasoning with Uncertainty. ECSQARU 2025 Proceedings Springer 2026 Cham 104 117 978-3-032-05133-2 Lecture Notes in Artificial Intelligence 16099 0302-9743 Sauerwald K. 340 Thimm M. 340