George Miloshevich
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
George Miloshevich has a total of 77 co-authors. Here are the top co-authors by number of shared publications:
- Stefaan Poedts
KU Leuven
7 shared publications - Panagiotis Gonidakis
Centre for mathematical Plasma Astrophysics (CmPA) KULeuven Leuven Belgium
5 shared publications - Philip J. Morrison
Department of Physics and Institute for Fusion Studies, The University of Texas at Austin 2 , Austin, Texas 78712
5 shared publications - Ekaterina Dineva
Centre for mathematical Plasma Astrophysics, Department of Mathematics, KU Leuven, Celestijnenlaan 200B, 3001 Leuven, Belgium;
4 shared publications Freddy Bouchet
ENSL, CNRS, Laboratoire de Physique, Lyon, France
4 shared publications- Felipe Nathan de Oliveira Lopes
Centre for Mathematical Plasma Astrophysics, Department of Mathematics, KU Leuven 1 , Celestijnenlaan 200B, 3001 Leuven
3 shared publications
Latest Publications
- Open Access
- 23 May 2026
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Machine Learning and Computation
- Vol. 3 (3)
Comparing Solar Structure Detection Methods in SDO/AIA Observations and the Application to Raw Uncalibrated Data
Recent advances in solar physics increasingly rely on automated identification of coronal structures using machine learning. Yet most studies emphasize scientific performance without evaluating feasibility for onboard deployment to prioritize downlink observations. This work investigates the automated identification of active regions and coronal holes by applying segmentation and detection techniques to Solar Dynamics Observatory (SDO) data. We compare three approaches: SCSS‐Net, a deep learning model for semantic segmentation; YOLOv8n, a lightweight object detector; and a traditional pipeline based on basic computer vision operations (BCVO). Each method is assessed for its scientific accuracy and its suitability for deployment in future resource‐limited missions. While no direct hardware benchmarking is performed in this study, we assess the feasibility of onboard implementation based on the associated number of trainable parameters, architecture and hardware requirements. Training and evaluation are first conducted on well‐calibrated SDO images. We then extend the evaluation to raw and uncalibrated SDO images affected by instrumental artifacts. Performance is measured using the Intersection over Union (IoU) and Dice score. Results show that while SCSS‐Net achieves the highest segmentation quality, YOLOv8n offers a strong balance between accuracy and efficiency. The BCVO pipeline remains viable under strict hardware limitations. Interestingly, our models retain compatibility on Level‐0 observations. This is the first study comparing these widely used methods from the perspective of onboard deployment. Our findings provide a foundation for designing frameworks tailored to onboard hardware configurations.
- 23 May 2026
- Published by American Geophysical Union (AGU) in Journal of Geophysical Research: Machine Learning and Computation
- Vol. 3 (3)
Open AccessComparing Solar Structure Detection Methods in SDO/AIA Observations and the Application to Raw Uncalibrated Data
Recent advances in solar physics increasingly rely on automated identification of coronal structures using machine learning. Yet most studies emphasize scientific performance without evaluating feasibility for onboard deployment to prioritize downlink observations. This work investigates the automated identification of active regions and coronal holes by applying segmentation and detection techniques to Solar Dynamics Observatory (SDO) data. We compare three approaches: SCSS‐Net, a deep learning model for semantic segmentation; YOLOv8n, a lightweight object detector; and a traditional pipeline based on basic computer vision operations (BCVO). Each method is assessed for its scientific accuracy and its suitability for deployment in future resource‐limited missions. While no direct hardware benchmarking is performed in this study, we assess the feasibility of onboard implementation based on the associated number of trainable parameters, architecture and hardware requirements. Training and evaluation are first conducted on well‐calibrated SDO images. We then extend the evaluation to raw and uncalibrated SDO images affected by instrumental artifacts. Performance is measured using the Intersection over Union (IoU) and Dice score. Results show that while SCSS‐Net achieves the highest segmentation quality, YOLOv8n offers a strong balance between accuracy and efficiency. The BCVO pipeline remains viable under strict hardware limitations. Interestingly, our models retain compatibility on Level‐0 observations. This is the first study comparing these widely used methods from the perspective of onboard deployment. Our findings provide a foundation for designing frameworks tailored to onboard hardware configurations.
- Open Access
- 14 March 2026
- Published by Copernicus GmbH
Probabilistic Solar Flare Forecasting via Weakly Supervised Contrastive Refinement of VAE Latent Spaces
Reliable solar flare forecasting is limited by two forms of class imbalance in active region time series: (i) the overwhelming dominance of the non-flaring, quiet state over the eruptive state, and (ii) the insufficient separability between common, physically similar event classes (e.g. C-class versus M-class flares). Although empirical parameters derived from the photospheric vector magnetic field (VMF), such as those provided by SDO/HMI SHARP products, capture aspects of active region complexity and free energy buildup, they often evolve smoothly and overlap across flare classes. Consequently, while many models can distinguish between flares and no-flares reasonably well, they struggle to distinguish flare magnitude and association with eruptive phenomena (e.g. CMEs) using photospheric information alone. This suggests that improved flare-class separation requires (a) the explicit definition of what constitutes 'similarity' between pre-flare states, and (b) parametrization that emphasizes flare-relevant structure over common active region features.We investigate a representation learning strategy that combines the parametrization of SDO/HMI SHARP VMF cutouts using a Variational Autoencoder (VAE) with a contrastive stage to reshape the resulting embedding geometry. First, a VAE is trained to encode SHARP cutouts into compact latent vectors that capture active region morphology. These vectors are then refined using a Siamese-like objective constructed from weak supervision, which uses event labels and empirical SHARP parameters as proxies for elevated flare likelihood. The contrastive stage then uses this weak supervision to encourage a latent geometry that better reflects flare-relevant evolution. This study emphasizes latent-space structure, i.e. neighborhood consistency and class-conditional clustering, and evaluates whether these properties facilitate improved probabilistic prediction across multiple forecast horizons, by training lightweight downstream models on (i) empirical parameters, (ii) VAE latents and (iii) their combined representations.
- 14 March 2026
- Published by Copernicus GmbH
Open AccessProbabilistic Solar Flare Forecasting via Weakly Supervised Contrastive Refinement of VAE Latent Spaces
Reliable solar flare forecasting is limited by two forms of class imbalance in active region time series: (i) the overwhelming dominance of the non-flaring, quiet state over the eruptive state, and (ii) the insufficient separability between common, physically similar event classes (e.g. C-class versus M-class flares). Although empirical parameters derived from the photospheric vector magnetic field (VMF), such as those provided by SDO/HMI SHARP products, capture aspects of active region complexity and free energy buildup, they often evolve smoothly and overlap across flare classes. Consequently, while many models can distinguish between flares and no-flares reasonably well, they struggle to distinguish flare magnitude and association with eruptive phenomena (e.g. CMEs) using photospheric information alone. This suggests that improved flare-class separation requires (a) the explicit definition of what constitutes 'similarity' between pre-flare states, and (b) parametrization that emphasizes flare-relevant structure over common active region features.We investigate a representation learning strategy that combines the parametrization of SDO/HMI SHARP VMF cutouts using a Variational Autoencoder (VAE) with a contrastive stage to reshape the resulting embedding geometry. First, a VAE is trained to encode SHARP cutouts into compact latent vectors that capture active region morphology. These vectors are then refined using a Siamese-like objective constructed from weak supervision, which uses event labels and empirical SHARP parameters as proxies for elevated flare likelihood. The contrastive stage then uses this weak supervision to encourage a latent geometry that better reflects flare-relevant evolution. This study emphasizes latent-space structure, i.e. neighborhood consistency and class-conditional clustering, and evaluates whether these properties facilitate improved probabilistic prediction across multiple forecast horizons, by training lightweight downstream models on (i) empirical parameters, (ii) VAE latents and (iii) their combined representations.
- Open Access
- 14 March 2026
- Published by Copernicus GmbH
Electron Neural Closure for Turbulent Magnetosheath Simulations
Modelling turbulence kinetically in space remains challenging due to the multiscale nature of plasma. An alternative approach is to adopt a fluid model hierarchy and close it using a phenomenological expression or law derived from local kinetic simulations. We address this challenge by examining decaying turbulence in the near-Earth magnetosheath using fully kinetic particle-in-cell (PIC) simulations [1]. We apply machine learning techniques to extract a non-local five-moment electron-pressure-tensor closure trained on these simulations. The data are carefully split across simulations initialized with different initial conditions, while maintaining the same turbulence and temperature levels. We evaluate the learned “equation of state” using energy-channel diagnostics, with emphasis on the pressure–strain interaction (a key mediator of turbulence heating). The new global closure outperforms common local approaches (e.g., double-adiabatic [2] and MLP-type closures [3]) in reconstructing key statistics. An equation of state trained on simulations with fewer particles per cell generalises to more accurate simulations with a higher number of particles per cell and different turbulent initialisations, while using the same physical parameters. Off-diagonal terms are more challenging to predict, but performance improves with the quantity of training data.Finally, we couple this data-driven electron closure with kinetic ion dynamics, advancing toward hybrid kinetic simulations in which electrons are represented by a neural network-based equation of state. This hybrid physics-informed machine learning framework offers a pathway to computationally efficient models with improved physical realism, potentially enabling both predictive simulations and parameter inference in heliospheric and magnetospheric applications.[1] G. Miloshevich, L. Vranckx, F.N. de Oliveira Lopes, P. Dazzi, G. Arrò, G. Lapenta, Phys. Plasmas 33 (2026) 012901.[2] A. Le, J. Egedal, W. Daughton, W. Fox, N. Katz, Phys. Rev. Lett. 102 (2009) 085001.[3] B. Laperre, J. Amaya, S. Jamal, G. Lapenta, Physics of Plasmas 29 (2022) 032706.
- 14 March 2026
- Published by Copernicus GmbH
Open AccessElectron Neural Closure for Turbulent Magnetosheath Simulations
Modelling turbulence kinetically in space remains challenging due to the multiscale nature of plasma. An alternative approach is to adopt a fluid model hierarchy and close it using a phenomenological expression or law derived from local kinetic simulations. We address this challenge by examining decaying turbulence in the near-Earth magnetosheath using fully kinetic particle-in-cell (PIC) simulations [1]. We apply machine learning techniques to extract a non-local five-moment electron-pressure-tensor closure trained on these simulations. The data are carefully split across simulations initialized with different initial conditions, while maintaining the same turbulence and temperature levels. We evaluate the learned “equation of state” using energy-channel diagnostics, with emphasis on the pressure–strain interaction (a key mediator of turbulence heating). The new global closure outperforms common local approaches (e.g., double-adiabatic [2] and MLP-type closures [3]) in reconstructing key statistics. An equation of state trained on simulations with fewer particles per cell generalises to more accurate simulations with a higher number of particles per cell and different turbulent initialisations, while using the same physical parameters. Off-diagonal terms are more challenging to predict, but performance improves with the quantity of training data.Finally, we couple this data-driven electron closure with kinetic ion dynamics, advancing toward hybrid kinetic simulations in which electrons are represented by a neural network-based equation of state. This hybrid physics-informed machine learning framework offers a pathway to computationally efficient models with improved physical realism, potentially enabling both predictive simulations and parameter inference in heliospheric and magnetospheric applications.[1] G. Miloshevich, L. Vranckx, F.N. de Oliveira Lopes, P. Dazzi, G. Arrò, G. Lapenta, Phys. Plasmas 33 (2026) 012901.[2] A. Le, J. Egedal, W. Daughton, W. Fox, N. Katz, Phys. Rev. Lett. 102 (2009) 085001.[3] B. Laperre, J. Amaya, S. Jamal, G. Lapenta, Physics of Plasmas 29 (2022) 032706.
- Open Access
- 14 March 2026
- Published by Copernicus GmbH
Data-Driven Identification of Region-Dependent Pressure Tensor Closures in Turbulent Space Plasmas
Understanding and modelling turbulence in space plasmas requires capturing kinetic effects that go beyond standard fluid closures. In the present work, we present a data-driven framework that combines unsupervised clustering and sparse equation discovery to identify effective closures in turbulent plasmas. Our primary focus is on solar-wind observations, but with possible applications to magnetospheric environments.We use unsupervised clustering methods, more specifically k-means, to identify dynamically similar regions in both in situ spacecraft data and numerical simulations. The first part of the project is focused on numerical simulations. Clustering is performed on multidimensional feature spaces constructed from plasma moments, fields, and other pressure-tensor-related quantities, applied to either 3D or 2D simulations. The resulting clusters define coherent regions characterized by comparable kinetic activity, anisotropy, and turbulence properties.These clustered regions serve as domains for sparse identification of nonlinear dynamics (SINDy). Particular emphasis is placed on exploring data-driven closures involving the pressure tensor, including anisotropic and nongyrotropic contributions, and understanding their role in momentum and other dynamical equations.The framework is designed to function consistently across both in situ measurements, such as Magnetospheric Multiscale (MMS) observations, and PIC simulations, enabling direct validation and comparison. This combined approach provides a structured method for discovering interpretable, region-specific closures in turbulent space plasmas and supports the development of reduced models directly informed by observations.
- 14 March 2026
- Published by Copernicus GmbH
Open AccessData-Driven Identification of Region-Dependent Pressure Tensor Closures in Turbulent Space Plasmas
Understanding and modelling turbulence in space plasmas requires capturing kinetic effects that go beyond standard fluid closures. In the present work, we present a data-driven framework that combines unsupervised clustering and sparse equation discovery to identify effective closures in turbulent plasmas. Our primary focus is on solar-wind observations, but with possible applications to magnetospheric environments.We use unsupervised clustering methods, more specifically k-means, to identify dynamically similar regions in both in situ spacecraft data and numerical simulations. The first part of the project is focused on numerical simulations. Clustering is performed on multidimensional feature spaces constructed from plasma moments, fields, and other pressure-tensor-related quantities, applied to either 3D or 2D simulations. The resulting clusters define coherent regions characterized by comparable kinetic activity, anisotropy, and turbulence properties.These clustered regions serve as domains for sparse identification of nonlinear dynamics (SINDy). Particular emphasis is placed on exploring data-driven closures involving the pressure tensor, including anisotropic and nongyrotropic contributions, and understanding their role in momentum and other dynamical equations.The framework is designed to function consistently across both in situ measurements, such as Magnetospheric Multiscale (MMS) observations, and PIC simulations, enabling direct validation and comparison. This combined approach provides a structured method for discovering interpretable, region-specific closures in turbulent space plasmas and supports the development of reduced models directly informed by observations.
- Open Access
- 1 January 2026
- Published by AIP Publishing in Physics of Plasmas
- Vol. 33 (1)
Electron neural closure for turbulent magnetosheath simulations: Energy channels
In this work, we introduce a non-local five-moment electron pressure tensor closure parametrized by a fully convolutional neural network (FCNN). Electron pressure plays an important role in generalized Ohm's law, competing with electron inertia. This model is used in the development of a surrogate model for a fully kinetic energy-conserving semi-implicit particle-in-cell simulation of decaying magnetosheath turbulence. We achieve this by training FCNN on a representative set of simulations with a smaller number of particles per cell and showing that our results generalize to a simulation with a large number of particles per cell. We evaluate the statistical properties of the learned equation of state, with a focus on pressure-strain interaction, which is crucial for understanding energy channels in turbulent plasmas. The resulting equation of state learned via FCNN significantly outperforms local closures, such as those learned by multi-layer perceptron (MLP) or double adiabatic expressions. We report that the overall spatial distribution of pressure-strain and its conditional averages are reconstructed well. However, some small-scale features are missed, especially for the off-diagonal components of the pressure tensor. Nevertheless, the results are substantially improved with more training data, indicating favorable scaling and potential for improvement, which will be addressed in future work.
- 1 January 2026
- Published by AIP Publishing in Physics of Plasmas
- Vol. 33 (1)
Open AccessElectron neural closure for turbulent magnetosheath simulations: Energy channels
In this work, we introduce a non-local five-moment electron pressure tensor closure parametrized by a fully convolutional neural network (FCNN). Electron pressure plays an important role in generalized Ohm's law, competing with electron inertia. This model is used in the development of a surrogate model for a fully kinetic energy-conserving semi-implicit particle-in-cell simulation of decaying magnetosheath turbulence. We achieve this by training FCNN on a representative set of simulations with a smaller number of particles per cell and showing that our results generalize to a simulation with a large number of particles per cell. We evaluate the statistical properties of the learned equation of state, with a focus on pressure-strain interaction, which is crucial for understanding energy channels in turbulent plasmas. The resulting equation of state learned via FCNN significantly outperforms local closures, such as those learned by multi-layer perceptron (MLP) or double adiabatic expressions. We report that the overall spatial distribution of pressure-strain and its conditional averages are reconstructed well. However, some small-scale features are missed, especially for the off-diagonal components of the pressure tensor. Nevertheless, the results are substantially improved with more training data, indicating favorable scaling and potential for improvement, which will be addressed in future work.
- Open Access
- 20 August 2025
- Vol. 989 (2)
Prediction of Time-evolving Radial Magnetic Fields on the Solar Surface Using Deep Learning
- 20 August 2025
- Vol. 989 (2)
Open AccessPrediction of Time-evolving Radial Magnetic Fields on the Solar Surface Using Deep Learning
- Open Access
- 18 March 2025
- Published by Copernicus GmbH
Fully kinetic and hybrid PIC modelling of magnetosheath turbulence: closure and dissipation
Understanding the interactions between the solar wind and the magnetosphere requires multi-scale modelling to resolve magnetohydrodynamic, ion and electron kinetic scales, owing to the collisionless character of plasma turbulence. This leads to computational complexity that reduced models aim to address.In this study, we investigate decaying turbulence in the magnetosheath by performing comparisons between the ECsim (a fully-kinetic Energy Conserving PIC) model and a computationally lighter model Menura (a hybrid PIC). Menura resolves kinetic ion scales but the influence of massless electrons is provided only via the pressure closure in the generalized Ohm’s law. To ensure meaningful comparisons, we have adjusted the initial conditions using parameters consistent with magnetosheath observations. We present a detailed analysis of the pressure-strain interaction terms, electromagnetic work and cross-scales fluxes, demonstrating relatively good agreement between the two models and validating certain turbulent characteristics for Menura. Our findings confirm several established results from fully kinetic Vlasov and PIC simulations, such as connections between coherent structures and energy conversion. Furthermore, we are extending these insights to a novel magnetosheath regime for a hybrid PIC model, which has generally received less attention in such studies. However, discrepancies were also identified, such as Zenitani measure (electromagnetic work done by the non-ideal electric field) and absolute values of energy dissipation which are model-dependent.In the effort to further improve electron pressure closure, we train a neural network surrogate on ECsim generated data (high fidelity model). We present preliminary results showing consistent scaling for predicted pressure-strain at future simulation time steps as a function of traceless stress, vorticity and the mean square total current density in a lack of data regime.
- 18 March 2025
- Published by Copernicus GmbH
Open AccessFully kinetic and hybrid PIC modelling of magnetosheath turbulence: closure and dissipation
Understanding the interactions between the solar wind and the magnetosphere requires multi-scale modelling to resolve magnetohydrodynamic, ion and electron kinetic scales, owing to the collisionless character of plasma turbulence. This leads to computational complexity that reduced models aim to address.In this study, we investigate decaying turbulence in the magnetosheath by performing comparisons between the ECsim (a fully-kinetic Energy Conserving PIC) model and a computationally lighter model Menura (a hybrid PIC). Menura resolves kinetic ion scales but the influence of massless electrons is provided only via the pressure closure in the generalized Ohm’s law. To ensure meaningful comparisons, we have adjusted the initial conditions using parameters consistent with magnetosheath observations. We present a detailed analysis of the pressure-strain interaction terms, electromagnetic work and cross-scales fluxes, demonstrating relatively good agreement between the two models and validating certain turbulent characteristics for Menura. Our findings confirm several established results from fully kinetic Vlasov and PIC simulations, such as connections between coherent structures and energy conversion. Furthermore, we are extending these insights to a novel magnetosheath regime for a hybrid PIC model, which has generally received less attention in such studies. However, discrepancies were also identified, such as Zenitani measure (electromagnetic work done by the non-ideal electric field) and absolute values of energy dissipation which are model-dependent.In the effort to further improve electron pressure closure, we train a neural network surrogate on ECsim generated data (high fidelity model). We present preliminary results showing consistent scaling for predicted pressure-strain at future simulation time steps as a function of traceless stress, vorticity and the mean square total current density in a lack of data regime.
- Open Access
- 18 March 2025
- Published by Copernicus GmbH
Parametrization of SHARP Vector Magnetic Field Using Disentangled Representation Learning
The rapid growth of high-dimensional data in solar physics presents significant challenges for analysis and interpretation, making it an excellent domain for the application of machine learning (ML) algorithms. Synoptic full-disk observations with the Solar Dynamics Observatory (SDO) provide continuous observations of the solar magnetic activity over more than one solar cycle, facilitating the study of solar variability and space weather impacts. The Space-weather HMI Active Region Patches (SHARP) vector magnetic field (VMF) maps and parameters, based on Helioseismic and Magnetic Imager (HMI) full-disk observations, are developed to study the magnetic evolution of individual active regions and flare triggering mechanisms. We present a method for active region parametrization by combining empirical parameters and ML-extracted features. Time series of SHARP VMF maps are used as input for the Disentangled Variational Autoencoder (VAE), a Disentangled Representation Learning (DRL) algorithm that facilitates the extraction of a low-dimensional feature representation. The VAE model is used to encode generalized information about nonlinear dynamical systems, i.e., a solar active region, aiming to isolate distinct factors of variation in the data, allowing a clearer interpretation of physical processes. We demonstrate how the ML features can be used to identify and study the stages of the magnetic patches evolution. These are benchmarked with SHARP parameters, relating empirical and learned features. Furthermore, the empirical dataset enhanced with ML features can be used to analyze the development of individual active regions and searching for eruption precursors.
- 18 March 2025
- Published by Copernicus GmbH
Open AccessParametrization of SHARP Vector Magnetic Field Using Disentangled Representation Learning
The rapid growth of high-dimensional data in solar physics presents significant challenges for analysis and interpretation, making it an excellent domain for the application of machine learning (ML) algorithms. Synoptic full-disk observations with the Solar Dynamics Observatory (SDO) provide continuous observations of the solar magnetic activity over more than one solar cycle, facilitating the study of solar variability and space weather impacts. The Space-weather HMI Active Region Patches (SHARP) vector magnetic field (VMF) maps and parameters, based on Helioseismic and Magnetic Imager (HMI) full-disk observations, are developed to study the magnetic evolution of individual active regions and flare triggering mechanisms. We present a method for active region parametrization by combining empirical parameters and ML-extracted features. Time series of SHARP VMF maps are used as input for the Disentangled Variational Autoencoder (VAE), a Disentangled Representation Learning (DRL) algorithm that facilitates the extraction of a low-dimensional feature representation. The VAE model is used to encode generalized information about nonlinear dynamical systems, i.e., a solar active region, aiming to isolate distinct factors of variation in the data, allowing a clearer interpretation of physical processes. We demonstrate how the ML features can be used to identify and study the stages of the magnetic patches evolution. These are benchmarked with SHARP parameters, relating empirical and learned features. Furthermore, the empirical dataset enhanced with ML features can be used to analyze the development of individual active regions and searching for eruption precursors.
- Open Access
- 18 March 2025
- Published by Copernicus GmbH
Efficient Segmentation and Clustering of Solar Coronal Structures: A Comparison of U-Net and Classical Computer Vision Techniques Using SDO Data
Segmentation and characterization of solar coronal structures are essential for advancing our understanding of solar atmosphere and accurately identifying key regions, such as active regions and coronal holes, which are precursors to phenomena like solar flares and coronal mass ejections (CMEs). In this study, we investigate two complementary approaches to automate this process. First, we employ a previously presented deep learning-based U-Net architecture tailored for segmenting and characterizing solar coronal structures. Second, we develop a lightweight algorithm aimed at optimizing resource efficiency, consisting of classical computer vision techniques, which include thresholding and morphological filtering. The approach that best balances segmentation performance and computational efficiency will be selected for integration into a prototype designed to support future space exploration missions.To characterize the segmented regions, we propose a set of carefully designed hand-crafted features to represent and characterize the resulting segmentations. These representations are analyzed using unsupervised clustering techniques, such as K-means and t-SNE, to distinguish solar coronal structures, including active regions, coronal holes and bright points.Our dataset spans multiple layers of the solar atmosphere, incorporating HMI magnetograms (photosphere) and AIA wavelengths—94 Å (flaring regions), 171 Å (quiet Sun), 193 Å (coronal structures), and 304 Å (chromosphere). The performance of both segmentation approaches is thoroughly evaluated using metrics such as Dice score and Intersection over Union (IoU), with comparisons made against state-of-the-art methods.Future work will focus on developing feature encoding techniques to better understand and predict solar phenomena, such as solar flare emissions, while investigating the impact of different feature extraction strategies on model performance.References:Galvez, Richard, et al. "A machine-learning data set prepared from the NASA solar dynamics observatory mission." The Astrophysical Journal Supplement Series 242.1 (2019): 7. Šimon Mackovjak et al. “SCSS-Net: solar corona structures segmentation by deep learning”, Monthly Notices of the Royal Astronomical Society, Volume 508, Issue 3, December 2021, Pages 3111–3124, https://doi.org/10.1093/mnras/stab2536 Gonidakis, Panagiotis & Sóñora-Mengana, Alexander & Jansen, Bart & Vandemeulebroucke, Jef. (2023). Handcrafted Features Can Boost Performance and Data-Efficiency for Deep Detection of Lung Nodules From CT Imaging. IEEE Access. PP. 1-1. 10.1109/ACCESS.2023.3331315.
- 18 March 2025
- Published by Copernicus GmbH
Open AccessEfficient Segmentation and Clustering of Solar Coronal Structures: A Comparison of U-Net and Classical Computer Vision Techniques Using SDO Data
Segmentation and characterization of solar coronal structures are essential for advancing our understanding of solar atmosphere and accurately identifying key regions, such as active regions and coronal holes, which are precursors to phenomena like solar flares and coronal mass ejections (CMEs). In this study, we investigate two complementary approaches to automate this process. First, we employ a previously presented deep learning-based U-Net architecture tailored for segmenting and characterizing solar coronal structures. Second, we develop a lightweight algorithm aimed at optimizing resource efficiency, consisting of classical computer vision techniques, which include thresholding and morphological filtering. The approach that best balances segmentation performance and computational efficiency will be selected for integration into a prototype designed to support future space exploration missions.To characterize the segmented regions, we propose a set of carefully designed hand-crafted features to represent and characterize the resulting segmentations. These representations are analyzed using unsupervised clustering techniques, such as K-means and t-SNE, to distinguish solar coronal structures, including active regions, coronal holes and bright points.Our dataset spans multiple layers of the solar atmosphere, incorporating HMI magnetograms (photosphere) and AIA wavelengths—94 Å (flaring regions), 171 Å (quiet Sun), 193 Å (coronal structures), and 304 Å (chromosphere). The performance of both segmentation approaches is thoroughly evaluated using metrics such as Dice score and Intersection over Union (IoU), with comparisons made against state-of-the-art methods.Future work will focus on developing feature encoding techniques to better understand and predict solar phenomena, such as solar flare emissions, while investigating the impact of different feature extraction strategies on model performance.References:Galvez, Richard, et al. "A machine-learning data set prepared from the NASA solar dynamics observatory mission." The Astrophysical Journal Supplement Series 242.1 (2019): 7. Šimon Mackovjak et al. “SCSS-Net: solar corona structures segmentation by deep learning”, Monthly Notices of the Royal Astronomical Society, Volume 508, Issue 3, December 2021, Pages 3111–3124, https://doi.org/10.1093/mnras/stab2536 Gonidakis, Panagiotis & Sóñora-Mengana, Alexander & Jansen, Bart & Vandemeulebroucke, Jef. (2023). Handcrafted Features Can Boost Performance and Data-Efficiency for Deep Detection of Lung Nodules From CT Imaging. IEEE Access. PP. 1-1. 10.1109/ACCESS.2023.3331315.
- Open Access
- 15 March 2025
- Published by Copernicus GmbH
Machine Learning Algorithms for Autonomous Space Mission Operations
The Automatics in SpAce exploration (ASAP) project has as a goal the design and development of Machine Learning algorithms for the automation of operations to be implemented on the on-board processors of space missions. In the framework of ASAP a set of ML algorithms for on-board science operations of space missions have been developed/optimized on consumer-grade computing systems to be further selected for orting of existent ML models directly on an FPGA prototype. In more detail, algorithms pertaining to four main use cases have been considered: the autonomous triggering of special measurement modes and the selective downlink of plasma environment parameters; the advanced on-board data analysis of three-dimensional particle distribution functions; the on-board analysis of solar images; the on-board prediction capability of SEP related hazards. Here we describe the algorithms, their performances and requirements for the on-board implementation. ASAP has received funding from the EU’s HORIZON Research and Innovation Action (GA no.101082633)
- 15 March 2025
- Published by Copernicus GmbH
Open AccessMachine Learning Algorithms for Autonomous Space Mission Operations
The Automatics in SpAce exploration (ASAP) project has as a goal the design and development of Machine Learning algorithms for the automation of operations to be implemented on the on-board processors of space missions. In the framework of ASAP a set of ML algorithms for on-board science operations of space missions have been developed/optimized on consumer-grade computing systems to be further selected for orting of existent ML models directly on an FPGA prototype. In more detail, algorithms pertaining to four main use cases have been considered: the autonomous triggering of special measurement modes and the selective downlink of plasma environment parameters; the advanced on-board data analysis of three-dimensional particle distribution functions; the on-board analysis of solar images; the on-board prediction capability of SEP related hazards. Here we describe the algorithms, their performances and requirements for the on-board implementation. ASAP has received funding from the EU’s HORIZON Research and Innovation Action (GA no.101082633)