More Like this

1. Partial-Attribution Instance Segmentation for Astronomical Source Detection and Deblending

   Hausen, R.; Robertson, B. (December 2021, Fourth Workshop on Machine Learning and the Physical Sciences (NeurIPS 2021))

   Astronomical source deblending is the process of separating the contribution of individual stars or galaxies (sources) to an image comprised of multiple, possibly overlapping sources. Astronomical sources display a wide range of sizes and brightnesses and may show substantial overlap in images. Astronomical imaging data can further challenge off-the-shelf computer vision algorithms owing to its high dynamic range, low signal-to-noise ratio, and unconventional image format. These challenges make source deblending an open area of astronomical research, and in this work, we introduce a new approach called Partial-Attribution Instance Segmentation that enables source detection and deblending in a manner tractable for deep learning models. We provide a novel neural network implementation as a demonstration of the method. 

   more » « less
2. Improving analogues-based detection & attribution approaches for hurricanes

   https://doi.org/10.1088/1748-9326/adaa8d  

   Bourdin, Stella; Camargo, Suzana_J; Lee, Chia-Ying; Lin, Jonathan; Vrac, Mathieu; Vaittinada_Ayar, Pradeebane; Faranda, Davide (January 2025, Environmental Research Letters)

   Abstract This paper presents a proof of concept for a new analogue-based framework for the detection and attribution of hurricane-related hazards. This framework addresses two important limitations of existing analogue-based methodologies: the lack of observed similar events, and the unsuitability of the distance metrics for hurricanes. To do so, we use a track-based metric, and we make use of synthetic tracks catalogues. We show that our method allows for selecting a sufficient number of suitable analogues, and we apply it to nine hurricane cases. Our analysis does not reveal any robust changes in wind hazards, translation speed, seasonality, or frequency over recent decades, consistent with current literature. This framework provides a reliable alternative to traditional analogue-based methods in the case of hurricanes, complementing and potentially enhancing efforts in addressing extreme weather event attribution. 

   more » « less
3. Data-Centric Machine Learning Approach for Early Ransomware Detection and Attribution

   https://doi.org/10.1109/NOMS56928.2023.10154378  

   Vehabovic, A.; Zanddizari, H.; Ghani, N.; Shaikh, F.; Bou-Harb, E.; Pour, M. Safaei; Crichigno, J. (May 2023, IEEE)
4. Toward Detection and Attribution of Cyber-Attacks in IoT-Enabled Cyber–Physical Systems

   https://doi.org/10.1109/JIOT.2021.3067667  

   Jahromi, Amir Namavar; Karimipour, Hadis; Dehghantanha, Ali; Choo, Kim-Kwang Raymond (September 2021, IEEE Internet of Things Journal)
5. Sensitivity of Seasonal Snowfall Attribution to Atmospheric Rivers and Their Reanalysis‐Based Detection

   https://doi.org/10.1029/2018GL080783  

   Huning, Laurie S.; Guan, Bin; Waliser, Duane E.; Lettenmaier, Dennis P. (January 2019, Geophysical Research Letters)

   Abstract We characterize the sensitivity of atmospheric river (AR)‐derived seasonal snowfall estimates to their atmospheric reanalysis‐based detection over Sierra Nevada, USA. We use an independent snow data set and the ARs identified with a single detection method applied to multiple atmospheric reanalyses of varying horizontal resolutions, to evaluate orographic relationships and contributions of individual ARs to the seasonal cumulative snowfall (CS). Spatial resolution differences have relatively minor effects on the number of ARs diagnosed, with higher‐resolution data sets identifying four more AR days per year, on average, during the 1985–2015 winters. However, this can lead to ~10% difference in AR attribution to the mean domain‐wide seasonal CS and differences up to 47% snowfall attribution at the seasonal scale. We show that identifying snow‐bearing ARs provides more information about the seasonal CS than simply knowing how many ARs occurred. Overall, we find that higher‐resolution atmospheric reanalyses imply greater attribution of seasonal CS to ARs. 

   more » « less