NNVub: a Neural Network Approach to $B\to X_u \ell \nu$

TL;DR

This paper employs neural networks to model shape functions in inclusive semileptonic B decays, enabling flexible incorporation of constraints and improving the extraction of |V_ub|, with implications for future Belle-II data analysis.

Contribution

It introduces a neural network method to parameterize shape functions in B decays, avoiding functional form assumptions and facilitating constraint implementation.

Findings

Good agreement with previous |V_ub| extractions

Neural network approach effectively incorporates experimental constraints

Discussion on future Belle-II data impact

Abstract

We use artificial neural networks to parameterize the shape functions in inclusive semileptonic $B$ decays without charm. Our approach avoids the adoption of functional form models and allows for a straightforward implementation of all experimental and theoretical constraints on the shape functions. The results are used to extract $|V_{ub}|$ in the GGOU framework and compared with the original GGOU paper and the latest HFAG results, finding good agreement in both cases. The possible impact of future Belle-II data on the $M_X$ distribution is also discussed.