Auto-annotation by ensemble of models is an efficient method of learning on unlabeled data. Wrong or inaccurate annotations generated by the ensemble may lead to performance degradation of the trained model. To deal with this problem we propose filte
ring the auto-labeled data using a trained model that predicts the quality of the annotation from the degree of consensus between ensemble models. Using semantic segmentation as an example, we show the advantage of the proposed auto-annotation filtering over training on data contaminated with inaccurate labels. Moreover, our experimental results show that in the case of semantic segmentation, the performance of a state-of-the-art model can be achieved by training it with only a fraction (30$%$) of the original manually labeled data set, and replacing the rest with the auto-annotated, quality filtered labels.
We set out to bridge the gap between regular black-hole spacetimes and observations of a black-hole shadow by the Event Horizon Telescope. We explore modifications of spinning and non-spinning black-hole spacetimes inspired by asymptotically safe qua
ntum gravity which features a scale dependence of the Newton coupling. As a consequence, the predictions of our model, such as the shadow shape and size, depend on one free parameter determining the curvature scale at which deviations from General Relativity set in. In more general new-physics settings, it can also depart substantially from the Planck scale. In this case, the free parameter is constrained by observations, since the corresponding curvature scale is significantly below the Planck-scale. The leading new-physics effect can be recast as a scale-dependent black-hole mass, resulting in distinct observational signatures of our model. As a concrete example, we show that two mass-measurements, extracted from the size of the shadow and from Keplerian orbital motion of stars, allow to distinguish the classical from the modified, regular black-hole spacetime, yielding a bound on the free parameter. For spinning black holes, we further find that the singularity-resolving new physics puts a characteristic dent in the shadow. Finally, we argue, based on the underlying physical mechanism, that the effects we derive could be generic consequences of a large class of quantum-gravity theories.
Gravitational wave and electromagnetic observations can provide new insights into the nature of matter at supra-nuclear densities inside neutron stars. Improvements in electromagnetic and gravitational wave sensing instruments continue to enhance the
accuracy with which they can measure the masses, radii, and tidal deformability of neutron stars. These better measurements place tighter constraints on the equation of state of cold matter above nuclear density. In this article, we discuss a complementary approach to get insights into the structure of neutron stars by providing a model prediction for non-linear fundamental eigenmodes (f-modes) and their decay over time, which are thought to be induced by time-dependent tides in neutron star binaries. Building on pioneering studies that relate the properties of f-modes to the structure of neutron stars, we systematically study this link in the non-perturbative regime using models that utilize numerical relativity. Using a suite of fully relativistic numerical relativity simulations of oscillating TOV stars, we establish blueprints for the numerical accuracy needed to accurately compute the frequency and damping times of f-mode oscillations, which we expect to be a good guide for the requirements in the binary case. We show that the resulting f-mode frequencies match established results from linear perturbation theory, but the damping times within numerical errors depart from linear predictions. This work lays the foundation for upcoming studies aimed at a comparison of theoretical models of f-mode signatures in gravitational waves, and their uncertainties with actual gravitational wave data, searching for neutron star binaries on highly eccentric orbits, and probing neutron star structure at high densities.
With the recent advent of multi-messenger gravitational-wave astronomy and in anticipation of more sensitive, next-generation gravitational-wave detectors, we investigate the dynamics, gravitational-wave emission, and nucleosynthetic yields of numero
us eccentric binary neutron-star mergers having different equations of state. For each equation of state we vary the orbital properties around the threshold of immediate merger, as well as the binary mass ratio. In addition to a study of the gravitational-wave emission including $f$-mode oscillations before and after merger, we couple the dynamical ejecta output from the simulations to the nuclear-reaction network code texttt{SkyNet} to compute nucleosynthetic yields and compare to the corresponding results in the case of a quasi-circular merger. We find that the amount and velocity of dynamically ejected material is always much larger than in the quasi-circular case, reaching maximal values of $M_{rm ej, max} sim 0.1 , M_{odot}$ and $v_{rm max}/c sim 0.75$. At the same time, the properties of this material are rather insensitive to the details of the orbit, such as pericenter distance or post-encounter apoastron distance. Furthermore, while the composition of the ejected matter depends on the orbital parameters and on the equation of state, the relative nucleosynthetic yields do not, thus indicating that kilonova signatures could provide information on the orbital properties of dynamically captured neutron-star binaries.
