Artificial intelligence and machine learning (AIML) techniques are widely used for recognising patterns in data and building models to describe complex relationships. Unlike traditional analytical models, machine learning (ML) models do not require an explicit theoretical formula in advance; instead, they learn the underlying trend from the data itself. This makes ML effective for problems involving nonlinear behaviour and multiple parameters, particularly in nuclear physics, where experimental information is often unavailable and theoretical models involve complex many body interactions. This work presents a preliminary exploration of ML based regression methods applied to β− decay systematics. In stellar environments or storage ring experiments in terrestrial laboratories, atoms may become fully ionised, allowing β− decay to unoccupied atomic orbitals (bound-state β− decay) as an additional channel alongside decay to the continuum. The relative importance of these channels depends sensitively on nuclear properties such as the decay Q value, mass number, and proton number of the daughter nucleus, making reliable prediction challenging. In this paper, ML regression models are trained on theoretically calculated decay rates for some nuclei to understand these dependencies and assess the potential of ML in uncovering systematic trends in weak interaction processes.
Aich et al. (2026) studied this question.