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