Pith. sign in

REVIEW 4 major objections 6 minor 56 references

Explainable AI for Solar Flare Prediction: Quantitative Magnetic Field Analysis of Model-Focused Regions

T0 review · 4 major / 6 minor · reviewed 2026-08-01 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Flare-prone active regions favor one magnetic polarity, AI maps show

desk verdict A competent quantitative XAI study whose central claim needs a random-mask baseline before it fully lands. read the letter →

arxiv 2607.15719 v1 pith:Y74APWK3 submitted 2026-07-17 astro-ph.SR

Z. Zheng , Q. Hao , C. Li , P.F. Chen , J.R. Hu , M.D. Ding , C. Fang This is my paper · ORCID
classification astro-ph.SR
keywords solarflarepredictionconvolutionalneuralnetworksexplainableAIGrad-CAMactiveregionsmagneticpolarityimbalancemagnetogramsspaceweather
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to show that a convolutional neural network trained to predict solar flares from magnetograms is not just a statistical black box: the regions it visually focuses on, isolated with the Grad-CAM attribution method, contain the same flare-relevant magnetic information that solar physicists extract from established active-region masks. The authors extract sixteen magnetic parameters from these model-focused regions (MFRs) and feed them to three standard classifiers, finding performance comparable to the standard full-region mask and better than a mask restricted to polarity inversion lines. They then define a polarity imbalance index and find that flare-producing active regions show an intermediate, single-polarity-dominated imbalance (roughly 1:2 positive-to-negative pixels), while quiet regions are more unipolar. The upshot is that deep-learning flare forecasts can be read as physical statements about magnetic complexity, making XAI a validation and discovery tool for solar physics, not just an explainability aid.

What carries the argument

Two constructions carry the argument. Model-focused regions (MFRs) are binary masks formed by applying OTSU thresholding to the absolute value of the Grad-CAM heatmap, where Grad-CAM weights are the global-average-pooled gradients of the flare score with respect to the last convolutional layer's feature maps; unlike standard Grad-CAM, the heatmap keeps both positive and negative activations, so the mask contains regions that promote or suppress flare predictions. The Polarity Imbalance Index (PII) is defined as |Np − Nn|/(Np + Nn) using pixels with |Blos| > 150 G inside each activation region, aggregated by area-weighting across connected components. MFRs localize what the network uses; PII

What would settle it

Ablation test: remove or zero the MFR mask pixels from input magnetograms and re-evaluate the CNN; if test-set performance does not degrade substantially relative to removing equal-area random regions, the MFRs are not the regions driving predictions and the physical interpretation collapses. A complementary check compares Grad-CAM masks with a different attribution method and requires high overlap.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the regions a CNN attends to when predicting flares are physically meaningful. Using Grad-CAM on the last convolutional layer and OTSU thresholding of the absolute heatmap to define model-focused regions (MFRs), the authors compute sixteen standard magnetic parameters inside those regions. Across three classifiers (random forest, SVM, fully connected network), MFR-derived parameters predict ≥C-class flares within 24 hours about as well as parameters from the full SHARP active-region mask, and clearly better than parameters from polarity-inversion-line masks. The second finding is a polarity signature: within positive-activation areas of

Load-bearing premise

The entire physical interpretation rests on Grad-CAM heatmaps faithfully marking the image regions the CNN actually uses; if the heatmaps highlight dataset shortcuts instead, the MFR parameters and polarity-imbalance results describe the model, not the Sun.

Editorial extensions

If this is right

  • MFR-derived magnetic parameters predict flares as well as parameters from the full active-region mask, so attention-based localization preserves the flare-relevant information in the magnetogram.
  • Polarity-inversion-line masks underperform because their limited spatial extent discards too much magnetic context; prediction systems should avoid relying on PIL strips alone.
  • Flare-productive active regions are characterized by intermediate polarity imbalance with one polarity dominating, giving a simple, observable discriminator for flare forecasting.
  • Grad-CAM's positive and negative activation regions separate cleanly by magnetic complexity, so CNN heatmaps can be decomposed into physical components rather than treated as opaque saliency.
  • If these results hold, CNN flare predictors can double as physical-discovery instruments, pointing to the specific magnetic configurations that precede flaring.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • One natural extension is to test PII as a standalone, line-of-sight-only flare predictor; if the distributional separation seen here holds, a simple pixel-count ratio could offer a lightweight operational complement to full network forecasts.
  • The roughly 1:2 dominant-polarity mixture may be a general property of flare-productive regions; checking it against independent active-region catalogs across multiple solar cycles would tell whether it is a stable physical signature or particular to this dataset.
  • A direct ablation — masking out MFR pixels and measuring how much CNN performance drops relative to random masks — would distinguish regions the network genuinely relies on from correlated bystanders; the paper does not report this test.
  • The difference between the leakage-free and leaky test splits (TSS 0.734 vs 0.762) implies that operational use on new active regions may underperform benchmark numbers, so chronological, region-disjoint evaluation deserves to become the default in this field.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. This paper proposes a quantitative XAI framework for CNN-based solar flare prediction. A ResNet-style CNN is trained on SDO/HMI SHARP vector magnetograms from 2699 HARPs with a two-stage, active-region-aware split. Grad-CAM (with both positive and negative activations retained) is used to define model-focused regions (MFRs) via OTSU thresholding of the absolute heatmap. The authors extract 16 standard SHARP magnetic parameters from MFR, SHARP, and PIL masks, and compare their predictive power in SVM, RF, and FCNN classifiers. They also introduce a Polarity Imbalance Index (PII) computed from LOS magnetogram pixels inside positive and negative activation areas, reporting that flare-productive regions have intermediate polarity imbalance in positive activation areas (PII_p ~ 0.33) while negative activation areas are more unipolar (PII_n ~ 1). The paper concludes that CNNs learn physically meaningful magnetic-complexity representations.

Significance. If the claims hold, the paper would demonstrate that XAI can move beyond qualitative heatmaps to quantitative physical validation, a valuable step for both solar physics and trustworthy AI. The authors are careful in several respects: they use active-region-based splitting to mitigate leakage, compare against the established SHARP parameter benchmark, use multiple heterogeneous classifiers with hyperparameter search, and make code and data publicly available. The PII construction is simple and directly tied to the CNN's attention. However, the central inference—that the specific regions highlighted by Grad-CAM are physically special—is not tested against a null model, and the PII analysis lacks uncertainty quantification. These are load-bearing gaps, because the paper's main conclusions rest on the selectivity of the MFR masks and on the statistical reality of the PII distribution differences.

major comments (4)
  1. [§4.1, Tables 6–7] The claim that MFR-derived parameters are 'comparable' to SHARP-derived parameters is not supported by uncertainty quantification. Table 6 reports mean metrics over hyperparameter configurations but no standard deviations, confidence intervals, or significance tests. Many differences are tiny (e.g., RF TSS: MFR 0.5408 vs SHARP 0.5543; FCNN F1: MFR 0.9067 vs SHARP 0.9079). Without paired tests across random seeds and configurations, 'comparable' is not established beyond visual inspection. Please report variability and perform significance tests on the metric differences.
  2. [§4.1, §2.2] There is no null control for the MFR mask itself. The SHARP mask covers the whole active region; any reasonably large, centrally located region will contain most of the magnetic flux and may yield similar parameter values. To show that Grad-CAM's selectivity matters, the parameter comparison should include random masks matched to the MFR masks in size, location, and contiguity (or a permutation/randomization of the Grad-CAM heatmap). Without such a control, the 'comparable predictive power' result is compatible with the MFR mask being a non-selective subsample, which would undermine the paper's central claim that the CNN has learned physically meaningful spatial features.
  3. [§4.2, Eq. (3), Fig. 6] The PII analysis is currently qualitative and based on several unvalidated choices. The paper reports KDE peaks (PII_p ~ 0.33, PII_n ~ 1) but provides no significance tests, confidence intervals, or effect sizes for the differences between positive and negative events. The definition of PII depends on an ad hoc ±150 G threshold, on OTSU thresholding of the absolute heatmap, on the sign convention of Grad-CAM activations, and on area-weighted averaging over connected components; none of these choices is tested for robustness. The sign of the Grad-CAM activation is assumed to indicate flare-promoting vs. suppressing regions without independent validation. Please add sensitivity analyses and formal statistical comparisons, and ideally validate the sign interpretation with an alternative attribution method or ablation.
  4. [§4.1, §2.1] The analysis of MFR predictive power is partly circular: the Grad-CAM masks are produced by a CNN trained on the same flare labels used to evaluate the MFR-derived parameters. The SHARP-mask comparison provides a useful external anchor, but it does not eliminate the possibility that the MFR parameters inherit predictive power from label-informed region selection. I recommend an additional control, e.g., evaluating parameters from Grad-CAM masks of an untrained or label-free CNN, or showing that MFR masks from a model trained on a different task do not yield comparable predictive power.
minor comments (6)
  1. [Title/Abstract] Line breaks in the title ('Model-F ocused') and several inline artifacts ('T able', 'V F ully', 'demostrate') should be fixed. The abstract's phrase 'magnetically complex configurations that are dominated by a single polarity' is slightly confusing given the later 'intermediately imbalanced' characterization; consider rewording for clarity.
  2. [References] Duplicate entries appear: Breiman (2001) twice, Schrijver (2007a/b) twice, and Li et al. (2025a) and (2025b) appear to be the same reference. Please consolidate.
  3. [§2, Data] The statement that the dataset comprises 'around half of all HARPs' is vague. Please give the exact selection criteria and the number of excluded HARPs due to central-meridian, NOAA-association, or NaN filtering.
  4. [§3.2, Hyperparameters] Table 5 lists 'random seed' as a hyperparameter, but the text says '60 different hyperparameter configurations' without explaining how the grid yields 60. Please clarify the counting.
  5. [§4.2, PII] The red/blue color coding in Figure 6 is described only in the caption; for accessibility, consider adding explicit labels or a colorblind-safe palette. Also, the caption says 'different solar active region events'—please specify whether each point is one timestamp or an aggregate over the active region's lifetime.
  6. [§5, Discussion] The discussion of PIL mask limitations is useful, but it would be strengthened by reporting the number of samples excluded in the 3-group vs 2-group alignments (i.e., how many HARPs lack PIL masks). Table 7 already shows the counts indirectly; please state them explicitly.

Circularity Check

1 steps flagged · score 2.0 of 10

Minor self-definitional overstatement; central validation rests on external SHARP comparison and held-out tests.

  1. self definitional [Section 4.1, first paragraph]
    "The CNN model achieved a true skill statistic (TSS) of 0.748 on the test set. This indicates that the model effectively discriminates between positive and negative samples rather than making random predictions. Thus, it is confirmed that the MFRs identified by the CNN serve as critical features for flare prediction."

    MFRs are defined by Eq. (1)-(2) as the regions that contribute most to the CNN's Grad-CAM output. Hence 'MFRs are critical features for the CNN's prediction' is true by construction once the CNN performs well; the sentence presents a definitional implication as an empirical confirmation. It does not by itself establish physical meaningfulness, since any accurate model's focused regions would automatically be 'critical' in this sense. This step is not load-bearing because the subsequent comparison to SHARP/PIL masks and the PII analysis provide external, non-circular validation.

full rationale

The paper's derivation chain is: train a CNN on flare labels; apply Grad-CAM (Eq. 1) and OTSU thresholding (Eq. 2) to obtain MFR masks; compute sixteen magnetic parameters inside MFR, SHARP, and PIL masks; train parameter-based classifiers on these features; and evaluate on a held-out test set that includes an active-region-disjoint partition (Test1). The SHARP and PIL masks are defined independently of the CNN (from BITMAP/CONF_DISAMBIG and Br thresholds, respectively), so the comparison of MFR-derived parameters against SHARP-derived parameters is an external benchmark. The test evaluation is out-of-sample: MFR masks for test samples come from a CNN that never saw those labels, and the parameter classifiers are trained on the training split. The PII analysis (Eq. 3) is an empirical description of the spatial distribution of polarities inside positive and negative activation areas; the observed difference between positive and negative events is not forced by the definitions. No load-bearing self-citations are used; the only author self-citation (P. F. Chen 2011) appears for background on space weather. The single tautological sentence in Section 4.1 is a presentational overreach, not the basis of the central claim. The absence of a random-mask null control is a correctness/experimental-design concern, but it is not circularity because the reported predictive power is not statistically forced by construction. Overall, the paper's central validation is self-contained and externally anchored; the circularity score is accordingly low.

Assumptions & free parameters 3 free parameters · 5 assumptions · 1 invented entities

The analysis leans on unvalidated modeling choices: Grad-CAM faithfulness, OTSU thresholding, PII threshold and pixel-count design, and dataset downsampling. The comparison to SHARP masks is the main external anchor.

free parameters (3)
  • PII polarity threshold = ±150 G
    Hand-chosen threshold in Eq. 3 to define positive/negative polarity pixels; directly controls PII and the paper's central complexity conclusion, with no sensitivity analysis.
  • CNN fixed input size = 512×512 pixels
    Chosen to standardize SHARP patches after zero-padding; the authors acknowledge in Sec. 5 that varying effective pixel scale may affect learned spatial features and introduce uncertainty.
  • Gaussian dilation sigma for PIL mask = σ = 10
    Adopted from Ran et al. (2022) for the PIL reference mask; not fitted here but affects the comparison baseline.
assumptions (5)
  • domain assumption Grad-CAM on the last convolutional layer faithfully identifies input regions driving the CNN's flare prediction.
    Central to MFR definition; not independently validated in the paper (Sec. 2.2, Eq. 1-2).
  • domain assumption OTSU thresholding of |Grad-CAM heatmap| yields a binary mask that captures the model's decision-relevant regions.
    Eq. 2; no validation that threshold choice preserves attribution structure.
  • ad hoc to paper PII computed from pixel counts above ±150 G is a valid measure of magnetic complexity relevant to flaring.
    Eq. 3; introduced here without external validation or sensitivity analysis.
  • domain assumption The downsampled dataset (all positives, one negative per AR per day) is representative of the operational flare-prediction problem.
    Sec. 2; may alter prevalence and active-region diversity.
  • standard math Standard ML classifiers (SVM/RF/FCNN) with the specified hyperparameter ranges adequately evaluate the predictive power of mask-derived parameters.
    Sec. 3.2; hyperparameters tuned on validation, but final evaluation on combined Test includes AR overlap.
invented entities (1)
  • Polarity Imbalance Index (PII)
    purpose: Quantify polarity balance inside Grad-CAM positive/negative activation regions from LOS magnetograms.
    New metric defined in Eq. 3; no independent calibration against established complexity measures; threshold and pixel-count choices are paper-specific.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Explainable AI for Solar Flare Prediction: Quantitative Magnetic Field Analysis of Model-Focused Regions." pith.science (2026). https://pith.science/paper/Y74APWK3

@misc{pith2026260715719,
  author       = {Pith},
  title        = {Pith review of: Explainable AI for Solar Flare Prediction: Quantitative Magnetic Field Analysis of Model-Focused Regions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y74APWK3}},
  note         = {Machine review of arXiv:2607.15719}
}
read the original abstract

Solar flares are intense energy release events in the solar atmosphere that may pose significant space weather hazards, which makes developing reliable prediction models essential. Although deep learning methods, particularly convolutional neural networks (CNNs), demonstrate strong predictive performance when using solar magnetograms, their scientific credibility is undermined by a lack of physical interpretability. Explainable artificial intelligence (XAI) offers a potential solution. However, current XAI studies in solar flare prediction are largely qualitative and lack systematic, theory-based, quantitative validation. We present a quantitative XAI framework that can decipher the physical basis of CNN-based solar flare prediction models. Using gradient-weighted class activation mapping (Grad-CAM), we identify model-focused regions (MFRs) in solar magnetograms. Then, we perform two key analyses to evaluate the predictive capability of magnetic parameters derived from MFRs and to quantitatively characterize their magnetic complexity. Our results reveal a strong physical correlation between MFRs and flare occurrence. Specifically, magnetic features extracted from MFRs demonstrate high predictive power for flares. Flare-producing active regions are characterized by magnetically complex configurations that are dominated by a single polarity rather than by balanced or purely unipolar structures. This finding is consistent with established physical theories of magnetic systems prone to flares. Our results suggest that CNNs can learn physically meaningful representations when trained on large-scale observations. Integrating XAI with quantitative magnetic field analysis improves the physical interpretability of deep learning-based flare prediction models, making them useful tools for prediction and modeling investigation in solar physics.

Figures

Figures reproduced from arXiv: 2607.15719 by the authors.

Figure 1
Figure 1. The flowchart of this study. It involves: (1) training a CNN-based flare prediction model using vector magnetic field data from solar active regions; (2) identifying MFRs via Grad-CAM; and (3) evaluating their physical significance through two quantitative experiments: comparing the predictive capability of parameter-based models using features extracted from MFR, SHARP, and PIL masks, and analyzing magnetic complex… view at source ↗
Figure 2
Figure 2. The flowchart of the data splitting. Test2 accounts for about 10% of the total dataset. We assessed the influence of active-region information leak￾age on performance estimation by comparing model per￾formance on Test1, which had no active-region overlap with the training set, and Test2, which allowed active￾region overlap. The detailed sample counts for each sub￾set are summarized in [PITH_FULL_IMAGE:figures/full_… view at source ↗
Figure 3
Figure 3. An example (HARP 5983 on 30 September 2015 at 03:00:00 UT) illustrating different mask types used in the analysis. (a) The line-of-sight magnetogram. (b) The Grad-CAM heatmap. (c) The overlapping region where BITMAP ≥ 30 (red) and CONF DISAMBIG ≥ 70 (white). (d) The PIL mask contours (green). (e) The MFR mask contours (blue). (f) The SHARP mask contours (red). in azimuthal disambiguation of the transverse magnetic f… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Schematic representation of the CNN model used in this work. Four-channel magnetic field images (Blos, Br, Bp, and Bt) with an input size (4, 512, 512) are processed through convolutional and residual blocks, followed by global average pooling and sigmoid classificatio…
Figure 5
Figure 5. Figure 5: Examples of active regions. (a) HARP 5983 on 30 September 2015 at 03:00:00 TAI. (b) HARP 2693 on 2 May 2013 at 08:00:00 TAI. (c) HARP 6167 on 10 December 2015 at 20:00:00 TAI. (d) HARP 2748 on 22 May 2013 at 08:00:00 TAI. (e) HARP 504 on 16 April 2011 at 08:00:00 TAI. …
Figure 6
Figure 6. Figure 6: The joint (center) and univariate (top and right) distributions of the Polarity Imbalance Index in positive and negative activation areas across different solar active regions. Each dot represents an active region event. The contours represent probability density: red …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 7 canonical work pages

  1. [1]

    Journal of Computational Physics , keywords =

    Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics , keywords =. doi:10.1016/j.jcp.2018.10.045 , adsurl =

    doi:10.1016/j.jcp.2018.10.045 2018

  2. [2]

    Proceedings of the 32nd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN 2024) , year =

    Leveraging Physics-Informed Neural Networks as Solar Wind Forecasting Models , author =. Proceedings of the 32nd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN 2024) , year =

    2024

  3. [3]

    , keywords =

    Solar Flare Prediction Using SDO/HMI Vector Magnetic Field Data with a Machine-learning Algorithm. , keywords =. doi:10.1088/0004-637X/798/2/135 , archivePrefix =. 1411.1405 , primaryClass =

    arXiv

  4. [4]

    Liu, Chang and Deng, Na and Wang, Jason T. L. and Wang, Haimin , year=. Predicting Solar Flares Using SDO/HMI Vector Magnetic Data Products and the Random Forest Algorithm , volume=. The Astrophysical Journal , publisher=. doi:10.3847/1538-4357/aa789b , number=

    doi:10.3847/1538-4357/aa789b

  5. [5]

    Living Reviews in Solar Physics , keywords =

    Machine learning in solar physics. Living Reviews in Solar Physics , keywords =. doi:10.1007/s41116-023-00038-x , archivePrefix =. 2306.15308 , primaryClass =

    arXiv

  6. [6]

    , keywords =

    Predicting Solar Flares Using a Long Short-term Memory Network. , keywords =. doi:10.3847/1538-4357/ab1b3c , archivePrefix =. 1905.07095 , primaryClass =

    arXiv 1905

  7. [7]

    Scientific Reports , year = 2023, month = aug, volume =

    Operational prediction of solar flares using a transformer-based framework. Scientific Reports , year = 2023, month = aug, volume =. doi:10.1038/s41598-023-40884-1 , adsurl =

    doi:10.1038/s41598-023-40884-1 2023

  8. [8]

    Deep Learning Based Solar Flare Forecasting Model. I. Results for Line-of-sight Magnetograms. , keywords =. doi:10.3847/1538-4357/aaae00 , adsurl =

    doi:10.3847/1538-4357/aaae00

Show all 56 references
  1. [9]

    , keywords =

    Solar Flare Forecast Using 3D Convolutional Neural Networks. , keywords =. doi:10.3847/1538-4357/ac9e53 , adsurl =

    doi

  2. [10]

    , keywords =

    Predicting Solar Flares Using CNN and LSTM on Two Solar Cycles of Active Region Data. , keywords =. doi:10.3847/1538-4357/ac64a6 , archivePrefix =. 2204.03710 , primaryClass =

    arXiv

  3. [11]

    , keywords =

    Parameters Derived from the SDO/HMI Vector Magnetic Field Data: Potential to Improve Machine-learning-based Solar Flare Prediction Models. , keywords =. doi:10.3847/1538-4357/ab441b , adsurl =

    doi

  4. [12]

    , keywords =

    Prediction of Large Solar Flares Based on SHARP and High-energy-density Magnetic Field Parameters. , keywords =. doi:10.3847/1538-4365/ad8b2a , archivePrefix =. 2410.18562 , primaryClass =

    arXiv

  5. [13]

    Science China Earth Sciences , keywords =

    Short-term solar eruptive activity prediction models based on machine learning approaches: A review. Science China Earth Sciences , keywords =. doi:10.1007/s11430-023-1375-2 , adsurl =

    doi

  6. [14]

    , keywords =

    Solar Flare Prediction Using Advanced Feature Extraction, Machine Learning, and Feature Selection. , keywords =. doi:10.1007/s11207-011-9896-1 , adsurl =

    doi

  7. [15]

    , keywords =

    The Helioseismic and Magnetic Imager (HMI) Vector Magnetic Field Pipeline: SHARPs - Space-Weather HMI Active Region Patches. , keywords =. doi:10.1007/s11207-014-0529-3 , archivePrefix =. 1404.1879 , primaryClass =

    arXiv

  8. [16]

    and Cogswell, Michael and Das, Abhishek and Vedantam, Ramakrishna and Parikh, Devi and Batra, Dhruv , booktitle=

    Selvaraju, Ramprasaath R. and Cogswell, Michael and Das, Abhishek and Vedantam, Ramakrishna and Parikh, Devi and Batra, Dhruv , booktitle=. Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization , year=

  9. [17]

    A Threshold Selection Method from Gray-Level Histograms , year=

    Otsu, Nobuyuki , journal=. A Threshold Selection Method from Gray-Level Histograms , year=

  10. [18]

    , keywords =

    Visual Explanation of a Deep Learning Solar Flare Forecast Model and Its Relationship to Physical Parameters. , keywords =. doi:10.3847/1538-4357/abdebe , adsurl =

    doi

  11. [19]

    Bobra and Xudong Sun and Michael J

    Monica G. Bobra and Xudong Sun and Michael J. Turmon , title =. doi:10.5281/zenodo.5131292 , url =

    doi

  12. [20]

    2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR , year = 2016, month = jun, eid =

    Deep Residual Learning for Image Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR , year = 2016, month = jun, eid =. doi:10.1109/CVPR.2016.90 , archivePrefix =. 1512.03385 , primaryClass =

    2016 arXiv

  13. [21]

    Journal of Machine Learning Research , year =

    Nitish Srivastava and Geoffrey Hinton and Alex Krizhevsky and Ilya Sutskever and Ruslan Salakhutdinov , title =. Journal of Machine Learning Research , year =

  14. [22]

    2014 , eprint=

    Network In Network , author=. 2014 , eprint=

    2014

  15. [23]

    , keywords =

    Identifying preflare spectral features using explainable artificial intelligence. , keywords =. doi:10.1051/0004-6361/202244835 , archivePrefix =. 2301.01560 , primaryClass =

    arXiv

  16. [24]

    , keywords =

    Correlation of the Coronal Mass Ejection Productivity of Solar Active Regions with Measures of Their Global Nonpotentiality from Vector Magnetograms: Baseline Results. , keywords =. doi:10.1086/339161 , adsurl =

    doi

  17. [25]

    , keywords =

    A Characteristic Magnetic Field Pattern Associated with All Major Solar Flares and Its Use in Flare Forecasting. , keywords =. doi:10.1086/511857 , adsurl =

    doi

  18. [27]

    Machine Learning , author =

    Support-vector networks , volume =. Machine Learning , author =. 1995 , pages =. doi:10.1007/BF00994018 , abstract =

    1995 doi

  19. [28]

    , year = 1986, month = oct, volume =

    Learning representations by back-propagating errors. , year = 1986, month = oct, volume =. doi:10.1038/323533a0 , adsurl =

    1986 doi

  20. [29]

    , keywords =

    Deep Flare Net (DeFN) Model for Solar Flare Prediction. , keywords =. doi:10.3847/1538-4357/aab9a7 , archivePrefix =. 1805.03421 , primaryClass =

    arXiv

  21. [30]

    , keywords =

    Relationship between Successive Flares in the Same Active Region and SHARP Parameters. , keywords =. doi:10.3847/1538-4357/ac80fa , archivePrefix =. 2207.07254 , primaryClass =

    arXiv

  22. [31]

    , keywords =

    Relationships between Characteristics of the Line-of-sight Magnetic Field and Solar Flare Forecasts. , keywords =. doi:10.3847/1538-4357/aa9119 , archivePrefix =. 1704.03423 , primaryClass =

    arXiv

  23. [32]

    Georgoulis, M. K. , title =. Geophysical Research Letters , volume =. 2007 , doi =

    2007

  24. [33]

    and DeVore, C

    Antiochos, Spiro K. and DeVore, C. Richard and Klimchuk, J. A. , title =. Astrophysical Journal , volume =. 1999 , doi =

    1999

  25. [34]

    2025 , eprint=

    PINN ME: A Physics-Informed Neural Network Framework for Accurate Milne-Eddington Inversions of Solar Magnetic Fields , author=. 2025 , eprint=

    2025

  26. [35]

    Living Reviews in Solar Physics , keywords =

    Solar Flares: Magnetohydrodynamic Processes. Living Reviews in Solar Physics , keywords =. doi:10.12942/lrsp-2011-6 , adsurl =

    2011 doi

  27. [36]

    astro-ph.SR

    An Observational Overview of Solar Flares. , primaryClass = "astro-ph.SR", keywords =. doi:10.1007/s11214-010-9701-8 , adsurl =

    doi

  28. [37]

    Living Reviews in Solar Physics , keywords =

    Coronal Mass Ejections: Models and Their Observational Basis. Living Reviews in Solar Physics , keywords =. doi:10.12942/lrsp-2011-1 , adsurl =

    2011 doi

  29. [38]

    Living Reviews in Solar Physics , keywords =

    Flare-productive active regions. Living Reviews in Solar Physics , keywords =. doi:10.1007/s41116-019-0019-7 , archivePrefix =. 1904.12027 , primaryClass =

    1904 arXiv

  30. [39]

    Science , keywords =

    A physics-based method that can predict imminent large solar flares. Science , keywords =. doi:10.1126/science.aaz2511 , adsurl =

    doi

  31. [40]

    , keywords =

    The Solar Dynamics Observatory (SDO). , keywords =. doi:10.1007/s11207-011-9841-3 , adsurl =

    doi

  32. [41]

    , keywords =

    Design and Ground Calibration of the Helioseismic and Magnetic Imager (HMI) Instrument on the Solar Dynamics Observatory (SDO). , keywords =. doi:10.1007/s11207-011-9842-2 , adsurl =

    doi

  33. [42]

    and Dumais, S.T

    Hearst, M.A. and Dumais, S.T. and Osuna, E. and Platt, J. and Scholkopf, B. , journal=. Support vector machines , year=

  34. [43]

    Leo Breiman , title =. Mach. Learn. , volume =. 2001 , url =. doi:10.1023/A:1010933404324 , timestamp =

    2001 doi

  35. [44]

    The Annals of Mathematical Statistics , number =

    Emanuel Parzen , title =. The Annals of Mathematical Statistics , number =. 1962 , doi =

    1962

  36. [45]

    The Annals of Mathematical Statistics , number =

    Murray Rosenblatt , title =. The Annals of Mathematical Statistics , number =. 1956 , doi =

    1956

  37. [46]

    , keywords =

    Solar Flare Forecasting Using Hybrid Neural Networks. , keywords =. doi:10.3847/1538-4365/ada281 , adsurl =

    doi

  38. [47]

    , keywords =

    Solar Flare Prediction Based on the Fusion of Multiple Deep-learning Models. , keywords =. doi:10.3847/1538-4365/ac249e , adsurl =

    doi

  39. [48]

    and Müller, Klaus-Robert , journal=

    Samek, Wojciech and Montavon, Grégoire and Lapuschkin, Sebastian and Anders, Christopher J. and Müller, Klaus-Robert , journal=. Explaining Deep Neural Networks and Beyond: A Review of Methods and Applications , year=

  40. [49]

    Universe , keywords =

    A Comprehensive Guide to Interpretable AI-Powered Discoveries in Astronomy. Universe , keywords =. doi:10.3390/universe11060187 , adsurl =

    doi

  41. [50]

    , keywords =

    Explaining deep learning of galaxy morphology with saliency mapping. , keywords =. doi:10.1093/mnras/stac368 , archivePrefix =. 2110.08288 , primaryClass =

    arXiv

  42. [51]

    Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI , journal =

    Alejandro. Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI , journal =. 2020 , issn =. doi:https://doi.org/10.1016/j.inffus.2019.12.012 , url =

    2020 doi

  43. [52]

    DBSCAN Revisited, Revisited: Why and How You Should (Still) Use DBSCAN , year =

    Schubert, Erich and Sander, J\". DBSCAN Revisited, Revisited: Why and How You Should (Still) Use DBSCAN , year =. doi:10.1145/3068335 , journal =

    doi

  44. [53]

    , keywords =

    The Helioseismic and Magnetic Imager (HMI) Investigation for the Solar Dynamics Observatory (SDO). , keywords =. doi:10.1007/s11207-011-9834-2 , adsurl =

    doi

  45. [54]

    , keywords =

    Forecasting Solar Flares Using Magnetogram-based Predictors and Machine Learning. , keywords =. doi:10.1007/s11207-018-1250-4 , archivePrefix =. 1801.05744 , primaryClass =

    arXiv

  46. [55]

    , keywords =

    Solar Flare Index Prediction Using SDO/HMI Vector Magnetic Data Products with Statistical and Machine-learning Methods. , keywords =. doi:10.3847/1538-4365/ac9b17 , archivePrefix =. 2209.13779 , primaryClass =

    arXiv

  47. [56]

    , keywords =

    Solar Flare Forecasting Using Machine Learning and SDO/HMI Data: A Multiple Machine Learning Model and Data Curation Technique Comparison Study. , keywords =. doi:10.3847/1538-4365/adf8e0 , adsurl =

    doi

  48. [57]

    , title =

    Nair, Vinod and Hinton, Geoffrey E. , title =. Proceedings of the 27th International Conference on International Conference on Machine Learning , pages =. 2010 , isbn =

    2010

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.