AM-IA brings together news, research papers and applications relating to artificial intelligence published around the world into a single feed — and makes them available in your European language at the click of a button.
Scroll to browse · Click to filter
By combining a Transformer on raw observations and a convolutive network on the periodogram, Hungarian and Swiss astronomers rank 37,006 sources without any annotated training data.
Astronomers have applied AdaCLIP, a "zero-shot" fault detection model designed outside astronomy, to the CERS and NGDEEP surveys to identify unusual morphologies.
A study compares Gaussian processes, recurring networks and a model with normalizing waves, FloTS, to anticipate the image quality of the sky, useful for telescopes as well as optical links in free space.
A two-layer convolutional neural network sorted through the dust impacts and solitary waves detected by Juno’s Waves instrument, revealing a ring of dust that is much more extensive than can be seen through a telescope.
Christian Kragh Jespersen defends the GNN as the most natural architecture for cosmology, while noting that they serve mainly on simulations.
Thought for spacecraft and future data centers in orbit, the Projection Healing method doubles the interval between two memory cleanings on commercial equipment.
By tracking atypical spectral distributions in 12 filters of the J-PLUS survey, astronomers isolate hundreds of rare candidates to be confirmed by spectroscopy.
Using automatic differentiation, a general method fits a photometric model to the observations and estimates the attitude of an object in orbit without the need for a dedicated model for each target.
An object detector adapted to damaged glass plates star and source markers to rescale their position on the sky, with a recovery rate of 97.3% and a two and a half times faster extraction.
Faced with the influx of data from large-scale sky surveys, a fully connected network and a SIREN model speed up the simultaneous fitting of light curves and radial velocity curves.
By applying simulation-based inference to the second CHIME catalogue (3,641 bursts), two researchers have concluded that there are indeed ‘unique’ sources, distinct from repetitive sources.
Since 2023, a central sensor on HiRISE has been malfunctioning intermittently, depriving the instrument of half its colour images. HiPredict fills the gap with an error margin of 1.1 per cent.
Tested in a simulation using Swift’s Burst Alert Telescope, an AI system directly deduces the positions of hard X-ray sources from detector counts, without knowing the pattern of the code-sliced image.
Parabolic jumps between waypoints, image segmentation to detect obstacles, and automatic retreat to a safe location: the framework has been tested in a high-fidelity lunar simulation.
Trained on five years’ worth of SuperDARN and DSCOVR data, three deep neural networks were compared: the probabilistic model performs better as the time horizon lengthens and the event intensifies.
The team responsible for operating the forecasting systems at the LBT and VLT observatories is testing neural networks against the autoregressive method, which is currently the benchmark for one-hour forecasts.
By simulating the pulsations of rotating δ Scuti stars, two astronomers have determined the age of the cluster from five stars observed by TESS, in line with conventional methods.
A new method applied to WASP-94A b utilises the temporal signature of limb asymmetry, reducing the error in the measured molecular abundances by a factor of 30 to 50.
An open library covers 48,037 polycyclic aromatic hydrocarbons across three charge states, with accuracy approaching that of quantum chemistry at a fraction of the cost.
Tested on the DP1 and DP2 data releases from the Vera C. Rubin Observatory, the pipeline achieves a macro-F1 of approximately 0.97, taking photometric uncertainties into account.