Abstract: The extraction of the neutrino mass ordering is one of the major challenges in particle physics and cosmology, not only for its implications for a fundamental theory of mass generation in nature, but also for its decisive role in the scale of future neutrinoless double beta decay experimental searches. It has been recently claimed that current oscillation, beta decay and cosmological limits on the different observables describing the neutrino mass parameter space provide robust decisive Bayesian evidence in favor of the normal ordering of the neutrino mass spectrum [1]. We further investigate these strong claims using a rich and wide phenomenology, with different sampling techniques of the neutrino parameter space. Contrary to the findings of Jimenez et al. [1], no decisive evidence for the normal mass ordering is found. Neutrino mass ordering analyses must rely on priors and parameterizations that are ordering-agnostic: robust results should be regarded as those in which the preference for the normal neutrino mass ordering is driven exclusively by the data, while we find a difference of up to a factor of 33 in the Bayes factors among the different priors and parameterizations exploited here. An ordering-agnostic prior would be represented by the case of parameterizations sampling over the two mass splittings and a mass scale, or those sampling over the individual neutrino masses via normal prior distributions only. In this regard, we show that the current significance in favor of the normal mass ordering should be taken as 2.7 sigma (i.e. moderate evidence), mostly driven by neutrino oscillation data. Let us stress that, while current data favor NO only mildly, we do not exclude the possibility that this may change in the future. Eventually, upcoming oscillation and cosmological data may (or may not) lead to a more significant exclusion of IO.

Abstract: The dark matter sector remains completely unknown. It is therefore crucial to keep an open mind regarding its nature and possible interactions. Focusing on the case of Weakly Interacting Massive Particles, in this work we make this general philosophy more concrete by applying modern machine learning techniques to dark matter direct detection. We do this by encoding and decoding the graphical representation of background events in the XENONnT experiment with a convolutional variational autoencoder. We describe a methodology that utilizes the `anomaly score' derived from the reconstruction loss of the convolutional variational autoencoder as well as a pre-trained standard convolutional neural network, in a semi-supervised fashion. Indeed, we observe that optimum results are obtained only when both unsupervised and supervised anomaly scores are considered together. A data set that has a higher proportion of anomaly score is deemed anomalous and deserves further investigation. Contrary to classical analyses, in principle all information about the events is used, preventing unnecessary information loss. Lastly, we demonstrate the reach of learning-focused anomaly detection in this context by comparing results with classical inference, observing that, if tuned properly, these techniques have the potential to outperform likelihood-based methods.