From RBMs to BN2A Models: Parameter Transformation for Interpretable Educational Diagnostics

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F25%3A10509003" target="_blank" >RIV/00216208:11320/25:10509003 - isvavai.cz</a>

Výsledek na webu

<a href="https://doi.org/10.1007/978-3-032-05134-9_8" target="_blank" >https://doi.org/10.1007/978-3-032-05134-9_8</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/978-3-032-05134-9_8" target="_blank" >10.1007/978-3-032-05134-9_8</a>

Jazyk výsledku

angličtina

Název v původním jazyce

From RBMs to BN2A Models: Parameter Transformation for Interpretable Educational Diagnostics

Popis výsledku v původním jazyce

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&apos;s interpretable parameters (leak and penalty probabilities), theoretical analysis showing BN2A&apos;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&apos;s representational power while achieving BN2A&apos;s interpretability, opening new possibilities for adaptive learning systems and diagnostic tools.

Název v anglickém jazyce

From RBMs to BN2A Models: Parameter Transformation for Interpretable Educational Diagnostics

Popis výsledku anglicky

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&apos;s interpretable parameters (leak and penalty probabilities), theoretical analysis showing BN2A&apos;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&apos;s representational power while achieving BN2A&apos;s interpretability, opening new possibilities for adaptive learning systems and diagnostic tools.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10102 - Applied mathematics

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Lecture Notes in Computer Science

ISBN

978-3-032-05134-9

ISSN

0302-9743

e-ISSN

1611-3349

Počet stran výsledku

14

Strana od-do

104-117

Název nakladatele

Springer Internat. Publ.

Místo vydání

Cham

Místo konání akce

Hagen, Germany

Datum konání akce

23. 9. 2025

Typ akce podle státní příslušnosti

EUR - Evropská akce

Kód UT WoS článku

—