pith. sign in

arxiv: 2605.29428 · v1 · pith:F23MJ65Hnew · submitted 2026-05-28 · 🌌 astro-ph.EP · astro-ph.IM· cs.AI

DELOS: Detecting Shallow Transits in Kepler Photometry Using a Contrastive-Learning Framework

Qingtian Liu , Jian Ge , XingChen Yan , Kevin Willis , Xinyu Yao , QuanQuan Hu , Jiapeng Zhu This is my paper

Pith reviewed 2026-06-29 01:07 UTC · model grok-4.3

classification 🌌 astro-ph.EP astro-ph.IMcs.AI
keywords transit detectionKepler photometrycontrastive learningexoplanet searchmachine learninglight curve analysisshallow transitsperiod search
0
0 comments X

The pith

A contrastive-learning framework improves detection of shallow transits in Kepler data by 15.5 percent over BLS in low-SNR conditions.

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

DELOS trains a convolutional encoder with contrastive learning on 20 million synthetic light curves to assign a transit-likeness score to each phase-folded curve and build a score periodogram. The approach focuses on intermediate-to-long periods of 100-150 days where shallow signals sit near the noise floor and traditional methods lose sensitivity. By scoring directly from folded data without pre-detected threshold events, the method raises combined precision-recall while cutting computation time. The paper demonstrates recovery of all known shallow signals in a validation sample and positions the tool for wider use on existing and upcoming photometric archives.

Core claim

DELOS uses GPU-accelerated phase folding and optimized binning together with a custom one-dimensional convolutional encoder trained contrastively on 20 million synthetic light curves that embed realistic transit models and Kepler-like noise; the encoder produces a transit-likeness score for each folded curve, yielding a periodogram that improves combined precision-recall by 15.5 percent relative to BLS and 11.25 percent relative to TLS specifically in the low-SNR regime, recovers every known shallow signal in the tested range, and accelerates the search by factors of 3-5 over BLS and 74-80 over TLS.

What carries the argument

A one-dimensional convolutional encoder trained via contrastive learning to score how closely an optimized phase-binned light curve matches a transit model.

If this is right

  • DELOS reaches 99.3 percent accuracy on a held-out synthetic validation set.
  • The framework recovers all known shallow intermediate-to-long-period transit signals present in the tested Kepler validation sample.
  • Search speed increases by factors of approximately 3-5 relative to BLS and 74-80 relative to TLS.
  • The same scoring approach is presented as directly extensible to transit searches in K2, TESS, PLATO, and Earth 2.0 photometry.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the synthetic training distribution matches real Kepler noise statistics, the encoder may surface additional shallow candidates that current pipelines miss.
  • Training the same contrastive encoder on periods longer than 150 days would provide a direct test of scalability toward terrestrial-planet detection.
  • Replacing threshold-crossing pre-filters with full score periodograms could reduce missed detections in large-scale surveys.

Load-bearing premise

The 20 million synthetic light curves with realistic transit models and Kepler-like noise properties capture the statistical features of actual Kepler photometry well enough for the encoder to generalize.

What would settle it

An injection-recovery experiment performed directly on real Kepler light curves in the 100-150 day range that shows DELOS recovering fewer confirmed low-SNR transits or producing lower combined precision-recall than BLS or TLS.

Figures

Figures reproduced from arXiv: 2605.29428 by Jian Ge, Jiapeng Zhu, Kevin Willis, Qingtian Liu, QuanQuan Hu, XingChen Yan, Xinyu Yao.

Figure 2
Figure 2. Figure 2: Conceptual diagram of DELOS consisting of four core components: the data augmentation Aug(·), the en￾coder network Enc(·), the projection network Proj(·), and the contrastive loss. Their relationship can be formulated as z = Proj(Enc(Aug(x))), where the source sample x can represent either transit or noise. The contrastive loss is em￾ployed to maximize the similarity between two augmented representations z… view at source ↗
Figure 1
Figure 1. Figure 1: Comparison of the periodograms generated by DELOS, BLS, and TLS on a simulated low-SNR, noise-nor￾malized light curve, each algorithm using the same trial pe￾riod sampling rate, Nsample = 58,400. The top panel shows the light curve folded at the correct 147.8-day transit period, highlighting the transit event with a localized zoom. The three panels below sequentially present the periodogram re￾sults from D… view at source ↗
Figure 3
Figure 3. Figure 3: Multi-view 3D t-SNE visualization of the projection-head representation for the validation set, shown from six combinations of viewing angles. Here, I denotes the input folded light curve, and φProj2(I) denotes the representation output by the second layer of the projection network. Blue points denote no-transit samples, and red points denote transit samples. The multiple viewpoints provide a more complete… view at source ↗
Figure 4
Figure 4. Figure 4: Schematic illustration of the DELOS framework. The framework consists of two stages: Representation Learn￾ing and Transit Detection. In the representation learning stage, augmented pairs generated from real transit and noise light-curve samples are used to optimize the encoder and projection network. Labels such as 4096-D and 2048-D de￾note the dimensionality of the corresponding input or latent feature ve… view at source ↗
Figure 5
Figure 5. Figure 5: End-to-end feature evolution and decision process of DELOS for two simulated examples: transit signal (top panel) and noise (bottom panel). The left column shows the folded-and-binned input light curve (4096 bins) and the feature maps of the five encoder convolution blocks, φci (I) (i = 1, . . . , 5), with channel-statistic traces (e.g., mean/max) summarizing layer-wise responses. The right column shows th… view at source ↗
Figure 6
Figure 6. Figure 6: Preprocessing of a real Kepler light curve and its changes before and after. We used a known ITL transit signal KIC 4138008 as the example. The top panel displays the raw PDCSAP light curve across 17 quarters, with the transit events marked in red. Dashed lines indicate the timing of each transit event, while the green line represents the best fitted smooth curve. The bottom panel presents the final light … view at source ↗
Figure 8
Figure 8. Figure 8: After obtaining these data pairs, they were randomly divided into three subsets: 80 percent for the training set, 10 percent for the validation set, and the remain￾ing 10 percent for the test set. The split was performed at the level of the original simulated source light curve, so that augmented views derived from the same source were never shared between the training, validation, and test sets. In order … view at source ↗
Figure 9
Figure 9. Figure 9: Comparative experimental results of DELOS, BLS, and TLS in terms of Receiver Operating Characteristic (ROC) and Area Under the Curve (AUC). The top panel shows the ROC curves for three methods at transit SNR ∈ {6, 7, 8, 11}. The bottom panel shows the AUC of the three methods within the transit SNR ∈ [6, 12]. DELOS achieves higher performance in distinguishing shallow transit from noise. to distinguish bet… view at source ↗
Figure 10
Figure 10. Figure 10: Comparative experimental results of DELOS, BLS, and TLS in terms of PR curve. The PR curves for the three methods at transit SNR ∈ {6, 7, 8, 11}, DELOS achieves higher precision and recall than BLS and TLS in the low-SNR regime, particularly for transit SNR below 11. (PR) curves were constructed, as shown in [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Based on the performance comparison among DELOS, BLS, and TLS, we draw the following conclusions. First, DELOS achieves the best overall performance in dis￾tinguishing shallow transit signals from noise. Second, DELOS shows higher sensitivity to shallow transit sig￾nals, as indicated by its higher TPR at a fixed FPR of 10 percent. Third, DELOS yields fewer false posi￾tives, as reflected by its lower FPR a… view at source ↗
Figure 12
Figure 12. Figure 12: Time cost of key step, UGPU, and ηSM during the contrastive learning inference stage vary with BatchSize. The solid lines represent the time cost, using the logarithmic scale on the left axis; the dashed lines represent GPU metrics, using the percentage scale on the right axis. bution, which can significantly enhance search efficiency. Consequently, to detect ITL low-SNR transits, we com￾pute the optimal … view at source ↗
Figure 14
Figure 14. Figure 14: DELOS score distributions for real Kepler sam￾ples and noise-control tests. The upper panel compares the scores of 3493 KOI non-transit samples with those of 60 known Kepler ITL, low-SNR transits, selected as described in Section 5.2. The red dashed line marks the adopted score threshold of 0.90. This threshold also recovers all 60 known ITL, low-SNR transit signals. The lower panel shows the score distri… view at source ↗
Figure 15
Figure 15. Figure 15: Example of a Kepler ITL exoplanetary signal recovered by DELOS. The upper panel presents the folded light curve of Kepler-1653 b at the correct transit period of 140.25 days, and the inset in the lower right corner provides a zoomed-in view of this transit. The lower panel displays the score periodogram produced by DELOS, revealing a promi￾nent peak at 140.25 days with 1.00 score, precisely aligning with … view at source ↗
Figure 16
Figure 16. Figure 16: Encoder network architecture of DELOS. The red cubes represent 1D Convolution layers (Conv1d) with ReLU activation, the blue cubes represent 1D Max Pooling layers (MaxPool1d) with BatchNorm layer (BN ) and Dropout layer (DP), and the gray cubes represent 1D Global Average Pooling layers (GlobalAvgPool1d). The numbers on the slanted edge of each cube indicate information related to the output. Specifically… view at source ↗
Figure 17
Figure 17. Figure 17: Variation of binned transit SNR with increasing binning number. The upper panel shows results for 4,000 synthetic light curves with original transit SNR ∈ [6, 8], while the lower panel corresponds to 6,000 synthetic light curves with original transit SNR ∈ [8, 11]. The red curves represent the binned transit SNR variations for each individual light curve. The yellow-shaded region marks the optimal binning… view at source ↗
Figure 18
Figure 18. Figure 18: Dependence of transit-duration and transit-depth preservation on the adopted phase-binning number. The relative error is defined as (binned value−original value)/original value×100%, with values closer to zero indicating better preservation of the original transit morphology after binning. The left panel shows the relative error in transit duration, while the right panel shows the relative error in transi… view at source ↗
Figure 19
Figure 19. Figure 19: Variations in DELOS loss and accuracy versus epochs during training. The bold red triangle highlights the epochs where DELOS achieved best performance. Trends in loss and accuracy throughout the training process were recorded, as shown in [PITH_FULL_IMAGE:figures/full_fig_p022_19.png] view at source ↗
read the original abstract

We present DEtection in phase-folded Light curves with cOntrastive Scoring (DELOS), a contrastive-learning-based framework designed to search for shallow transits in Kepler photometry. DELOS combines GPU-accelerated phase folding, optimized phase binning, and a custom one-dimensional convolutional encoder to assign a transit-likeness score to each folded light curve, thereby producing a score periodogram over trial periods without relying on pre-detected threshold-crossing events. Focusing on intermediate-to-long-period signals with orbital periods of 100-150 days, DELOS was trained on 20 million synthetic light curves generated with realistic transit models and Kepler-like noise properties, achieving a validation accuracy of 99.3 percent on the synthetic validation set. In controlled injection-recovery experiments, DELOS improves the combined precision-recall performance by 15.5 percent relative to Box-fitting Least Squares (BLS) and 11.25 percent relative to Transit Least Squares (TLS) in the low Signal-to-Noise Ratios (low-SNR) regime. It also accelerates the search by factors of approximately 3-5 and 74-80 compared with BLS and TLS, respectively. Applied to a selected Kepler validation sample, DELOS recovered all known shallow intermediate-to-long-period transit signals in the tested period range. These results demonstrate that DELOS provides an efficient and sensitive framework for low-SNR transit searches and represents a practical step toward future searches for longer-period terrestrial planets in Kepler, K2, TESS, PLATO, and Earth 2.0 data. Accordingly, this work is intended as a methodological development and validation study, with the detailed astrophysical validation of newly identified candidates deferred to future work.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces DELOS, a contrastive-learning framework using a 1D convolutional encoder for detecting shallow transits in Kepler photometry at periods of 100-150 days. Trained on 20 million synthetic light curves with realistic transit models and Kepler-like noise, it reports 99.3% validation accuracy on synthetics. Injection-recovery tests show 15.5% and 11.25% gains in combined precision-recall over BLS and TLS in the low-SNR regime, with speedups of ~3-5x and 74-80x. On a selected Kepler validation sample, it recovers all known shallow intermediate-to-long-period signals. The work positions itself as a methodological study deferring astrophysical validation of new candidates.

Significance. If the generalization from synthetics holds, DELOS could provide an efficient tool for low-SNR transit searches in Kepler and future missions like PLATO. Strengths include the scale of the synthetic training set, direct quantitative comparisons to BLS/TLS, reported computational accelerations, and explicit recovery of known real signals. The contrastive-learning approach and GPU-accelerated phase folding represent a practical methodological contribution for handling shallow transits.

major comments (2)
  1. [Abstract / real-data validation] Abstract and results on real-data application: recovery of all known signals in the selected Kepler validation sample is reported, but no quantitative metrics (e.g., precision-recall, false-positive rate, or detection efficiency) are provided for the real photometry itself. This leaves the claimed performance gains (15.5%/11.25%) supported only by synthetic injection-recovery and weakens the assertion that DELOS is a 'sensitive framework' for actual Kepler data.
  2. [Methods / synthetic data generation] Methods / training data description: the central generalization claim rests on the 20 million synthetic light curves capturing 'Kepler-like noise properties' for the 100-150 day range, yet no explicit tests or metrics are given for how well non-stationary systematics, pixel correlations, or gap patterns in real Kepler data are reproduced by the noise model.
minor comments (2)
  1. [Results / injection-recovery experiments] Specify the exact SNR thresholds or range used to define the 'low-SNR regime' in the injection-recovery comparisons.
  2. [Methods] The phase binning scheme and 1D convolutional encoder hyperparameters are listed as free parameters; providing their specific values or ranges would improve reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the positive assessment and recommendation of minor revision. We address each major comment below with honest responses and indicate planned changes to the manuscript.

read point-by-point responses

  1. Referee: [Abstract / real-data validation] Abstract and results on real-data application: recovery of all known signals in the selected Kepler validation sample is reported, but no quantitative metrics (e.g., precision-recall, false-positive rate, or detection efficiency) are provided for the real photometry itself. This leaves the claimed performance gains (15.5%/11.25%) supported only by synthetic injection-recovery and weakens the assertion that DELOS is a 'sensitive framework' for actual Kepler data.

    Authors: We agree that the absence of quantitative metrics on real photometry limits the strength of claims about performance on actual Kepler data. The real-data test was designed as a qualitative sanity check confirming recovery of all known signals in a curated sample, consistent with the paper's framing as a methodological study that defers full astrophysical validation. We will revise the abstract, results, and discussion sections to explicitly state that performance gains are quantified only on synthetics, clarify the scope of the real-data test, and add available details such as sample size and any observed false-positive behavior where possible. A complete precision-recall analysis on unlabeled real data is not feasible without additional ground-truth efforts outside this work's scope. revision: partial

  2. Referee: [Methods / synthetic data generation] Methods / training data description: the central generalization claim rests on the 20 million synthetic light curves capturing 'Kepler-like noise properties' for the 100-150 day range, yet no explicit tests or metrics are given for how well non-stationary systematics, pixel correlations, or gap patterns in real Kepler data are reproduced by the noise model.

    Authors: The synthetic noise is constructed from established Kepler noise models in the literature, including components for correlated noise and observational gaps drawn from real data distributions. We acknowledge that explicit quantitative validation (e.g., power-spectrum comparisons or gap statistics) is not currently presented. We will expand the methods section with additional description of the noise-generation procedure and any internal consistency checks performed during dataset creation. Adding new comparative figures or metrics would require further analysis not present in the current manuscript. revision: yes

Circularity Check

0 steps flagged

No circularity; performance metrics independent of model parameters

full rationale

The paper's central claims rest on injection-recovery tests using held-out synthetic light curves (20M training set, separate validation) compared against independent standard algorithms BLS and TLS, plus recovery of known real signals. These quantities are not defined in terms of the DELOS encoder's outputs or fitted parameters, nor do they rely on self-citations or ansatzes imported from prior author work. The derivation chain is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The central performance claims rest on the fidelity of the synthetic training distribution to real Kepler noise and on the choice of convolutional encoder architecture and binning parameters, none of which receive independent external validation in the abstract.

free parameters (2)
  • phase binning scheme
    Optimized phase binning is part of the pipeline and chosen to enhance transit contrast.
  • 1D convolutional encoder hyperparameters
    Custom encoder architecture and training details are tuned on the synthetic set.
axioms (1)
  • domain assumption Synthetic light curves with Kepler-like noise accurately represent the statistical properties of real photometry for 100-150 day periods.
    Training, validation accuracy, and injection-recovery tests all depend on this match.

pith-pipeline@v0.9.1-grok · 5870 in / 1510 out tokens · 52221 ms · 2026-06-29T01:07:00.621511+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

80 extracted references · 77 canonical work pages · 6 internal anchors

  1. [1]

    2015, TensorFlow: Large-Scale Machine Learning on Heterogeneous Systems, https://www.tensorflow.org/

    Abadi, M., Agarwal, A., Barham, P., et al. 2015, TensorFlow: Large-Scale Machine Learning on Heterogeneous Systems, https://www.tensorflow.org/

    2015

  2. [2]

    P., Fouqu´ e, P., et al

    Beaulieu, J.-P., Bennett, D. P., Fouqu´ e, P., et al. 2006, Nature, 439, 437, doi: 10.1038/nature04441

    work page doi:10.1038/nature04441 2006

  3. [3]

    J., Koch, D., Basri, G., et al

    Borucki, W. J., Koch, D., Basri, G., et al. 2010, Science, 327, 977, doi: 10.1126/science.1185402

    work page doi:10.1126/science.1185402 2010

  4. [4]

    J., Koch, D

    Borucki, W. J., Koch, D. G., Batalha, N., et al. 2012, The Astrophysical Journal, 745, 120, doi: 10.1088/0004-637X/745/2/120

    work page doi:10.1088/0004-637x/745/2/120 2012

  5. [5]

    C., Emilio, M., Pacheco, E

    Boufleur, R. C., Emilio, M., Pacheco, E. J., de La Reza, J. R., & da Rocha, J. C. 2014, in IAU Symposium, Vol. 293, Formation, Detection, and Characterization of Extrasolar Habitable Planets, ed. N. Haghighipour, 410–412, doi: 10.1017/S1743921313013288

    work page doi:10.1017/s1743921313013288 2014

  6. [6]

    Bromley, J., Guyon, I., LeCun, Y., S¨ ackinger, E., & Shah, R. 1993, in Proceedings of the 6th International Conference on Neural Information Processing Systems, NIPS’93 (San Francisco, CA, USA: Morgan Kaufmann Publishers Inc.), 737–744, doi: 10.5555/2987189.2987282

    work page doi:10.5555/2987189.2987282 1993

  7. [7]

    J., Christiansen, J

    Burke, C. J., Christiansen, J. L., Mullally, F., et al. 2015, The Astrophysical Journal, 809, 8, doi: 10.1088/0004-637X/809/1/8

    work page doi:10.1088/0004-637x/809/1/8 2015

  8. [8]

    Campbell, B., Walker, G. A. H., & Yang, S. 1988, The Astrophysical Journal, 331, 902, doi: 10.1086/166608

    work page doi:10.1086/166608 1988

  9. [9]

    A., & Agol, E

    Carter, J. A., & Agol, E. 2013, The Astrophysical Journal, 765, 132, doi: 10.1088/0004-637X/765/2/132

    work page doi:10.1088/0004-637x/765/2/132 2013

  10. [10]

    M., Latham, D

    Charbonneau, D., Brown, T. M., Latham, D. W., & Mayor, M. 2000, The Astrophysical Journal, 529, L45, doi: 10.1086/312457

    work page doi:10.1086/312457 2000

  11. [11]

    R., et al

    Chaushev, A., Raynard, L., Goad, M. R., et al. 2019, Monthly Notices of the Royal Astronomical Society, 488, 5232, doi: 10.1093/mnras/stz2058

    work page doi:10.1093/mnras/stz2058 2019

  12. [12]

    2004, Astronomy & Astrophysics, 425, L29, doi: 10.1051/0004-6361:200400056

    Chauvin, G., Lagrange, A.-M., Dumas, C., et al. 2004, Astronomy & Astrophysics, 425, L29, doi: 10.1051/0004-6361:200400056

    work page doi:10.1051/0004-6361:200400056 2004

  13. [13]

    doi: 10.1109/ICCV48922.2021.00951

    Chen, X., Xie, S., & He, K. 2021, in 2021 IEEE/CVF International Conference on Computer Vision (ICCV), 9620–9629, doi: 10.1109/ICCV48922.2021.00950

    work page doi:10.1109/iccv48922.2021.00950 2021

  14. [14]

    2019, Publications of the Astronomical Society of the Pacific, 131, 064502, doi: 10.1088/1538-3873/ab13d3

    Chintarungruangchai, P., & Jiang, I.-G. 2019, Publications of the Astronomical Society of the Pacific, 131, 064502, doi: 10.1088/1538-3873/ab13d3

    work page doi:10.1088/1538-3873/ab13d3 2019

  15. [15]

    2005, in 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05), Vol

    Chopra, S., Hadsell, R., & LeCun, Y. 2005, in 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05), Vol. 1, 539–546 vol. 1, doi: 10.1109/CVPR.2005.202

    work page doi:10.1109/cvpr.2005.202 2005

  16. [16]

    2011, Astronomy & Astrophysics, 529, A75, doi: 10.1051/0004-6361/201116451

    Claret, A., & Bloemen, S. 2011, Astronomy & Astrophysics, 529, A75, doi: 10.1051/0004-6361/201116451

    work page doi:10.1051/0004-6361/201116451 2011

  17. [17]

    The Astropy Project: Sustaining and Growing a Community-oriented Open-source Project and the Latest Major Release (v5.0) of the Core Package

    Collaboration, A., Price-Whelan, A. M., Lim, P. L., et al. 2022, The Astrophysical Journal, 935, 167, doi: 10.3847/1538-4357/ac7c74 Collier Cameron, A., Pollacco, D., Street, R. A., et al. 2006, Monthly Notices of the Royal Astronomical Society, 373, 799, doi: 10.1111/j.1365-2966.2006.11074.x

    work page internal anchor Pith review doi:10.3847/1538-4357/ac7c74 2022

  18. [18]

    2021, The Astronomical Journal, 163, 23, doi: 10.3847/1538-3881/ac3482

    Cui, K., Liu, J., Feng, F., & Liu, J. 2021, The Astronomical Journal, 163, 23, doi: 10.3847/1538-3881/ac3482

    work page doi:10.3847/1538-3881/ac3482 2021

  19. [19]

    J., et al

    Dattilo, A., Vanderburg, A., Shallue, C. J., et al. 2019, The Astronomical Journal, 157, 169, doi: 10.3847/1538-3881/ab0e12

    work page doi:10.3847/1538-3881/ab0e12 2019

  20. [20]

    D., & Charbonneau, D

    Dressing, C. D., & Charbonneau, D. 2013, The Astrophysical Journal, 767, 95, doi: 10.1088/0004-637X/767/1/95

    work page doi:10.1088/0004-637x/767/1/95 2013

  21. [21]

    D., & Charbonneau, D

    Dressing, C. D., & Charbonneau, D. 2015, The Astrophysical Journal, 807, 45, doi: 10.1088/0004-637X/807/1/45

    work page doi:10.1088/0004-637x/807/1/45 2015

  22. [22]

    W., & Morton, T

    Foreman-Mackey, D., Hogg, D. W., & Morton, T. D. 2014, The Astrophysical Journal, 795, 64, doi: 10.1088/0004-637X/795/1/64

    work page doi:10.1088/0004-637x/795/1/64 2014

  23. [23]

    2013, The Astrophysical Journal, 766, 81, doi: 10.1088/0004-637X/766/2/81 24Liu et al

    Fressin, F., Torres, G., Charbonneau, D., et al. 2013, The Astrophysical Journal, 766, 81, doi: 10.1088/0004-637X/766/2/81 24Liu et al

    work page doi:10.1088/0004-637x/766/2/81 2013

  24. [24]

    1980, Biological Cybernetics, 36, 193, doi: 10.1007/BF00344251

    Fukushima, K. 1980, Biological Cybernetics, 36, 193, doi: 10.1007/BF00344251

    work page doi:10.1007/bf00344251 1980

  25. [25]

    Ge, J., Zhang, H., Deng, H., & Howell, S. B. 2022a, The Innovation, 3, doi: 10.1016/j.xinn.2022.100271

    work page doi:10.1016/j.xinn.2022.100271 2022

  26. [26]

    2022b, arXiv:2206.06693, doi: https://doi.org/10.48550/arXiv.2206.06693

    Ge, J., Zhang, H., Zand, W., & Deng, H. 2022b, arXiv:2206.06693, doi: https://doi.org/10.48550/arXiv.2206.06693

    work page doi:10.48550/arxiv.2206.06693

  27. [27]

    2022c, in Space Telescopes and Instrumentation 2022: Optical, Infrared, and Millimeter Wave, ed

    Ge, J., Zhang, H., Zhang, Y., et al. 2022c, in Space Telescopes and Instrumentation 2022: Optical, Infrared, and Millimeter Wave, ed. L. E. Coyle, S. Matsuura, & M. D. Perrin, Vol. 12180, International Society for Optics and Photonics (SPIE), 1218015, doi: 10.1117/12.2630656

    work page doi:10.1117/12.2630656 2022

  28. [28]

    2024a, Chinese Journal of Space Science, 44, 400, doi: 10.11728/cjss2024.03.yg05

    Ge, J., Chen, W., Chen, Y., et al. 2024a, Chinese Journal of Space Science, 44, 400, doi: 10.11728/cjss2024.03.yg05

    work page doi:10.11728/cjss2024.03.yg05

  29. [29]

    2024b, in Space Telescopes and Instrumentation 2024: Optical, Infrared, and Millimeter Wave, ed

    Ge, J., Zhang, H., Zhang, Y., et al. 2024b, in Space Telescopes and Instrumentation 2024: Optical, Infrared, and Millimeter Wave, ed. L. E. Coyle, S. Matsuura, & M. D. Perrin, Vol. 13092, International Society for Optics and Photonics (SPIE), 1309218, doi: 10.1117/12.3018669

    work page doi:10.1117/12.3018669 2024

  30. [30]

    2020, in Proceedings of the 34th International Conference on Neural Information Processing Systems, NIPS ’20 (Red

    Grill, J.-B., Strub, F., Altch´ e, F., et al. 2020, in Proceedings of the 34th International Conference on Neural Information Processing Systems, NIPS ’20 (Red

    2020

  31. [31]

    Hook, NY, USA: Curran Associates Inc.), doi: 10.5555/3495724.3497510

    work page doi:10.5555/3495724.3497510

  32. [32]

    Falcon: A fourier transform based approach for fast and secure convolutional neural network predictions

    He, K., Fan, H., Wu, Y., Xie, S., & Girshick, R. 2020, in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 9726–9735, doi: 10.1109/CVPR42600.2020.00975

    work page doi:10.1109/cvpr42600.2020.00975 2020

  33. [33]

    2015, in 2015 IEEE International Conference on Computer Vision (ICCV), 1026–1034, doi: 10.1109/ICCV.2015.123

    He, K., Zhang, X., Ren, S., & Sun, J. 2015, in 2015 IEEE International Conference on Computer Vision (ICCV), 1026–1034, doi: 10.1109/ICCV.2015.123

    work page doi:10.1109/iccv.2015.123 2015

  34. [34]

    Deep Residual Learning for Image Recognition

    He, K., Zhang, X., Ren, S., & Sun, J. 2016, in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 770–778, doi: 10.1109/CVPR.2016.90

    work page doi:10.1109/cvpr.2016.90 2016

  35. [35]

    W., Marcy, G

    Henry, G. W., Marcy, G. W., Butler, R. P., & Vogt, S. S. 2000, The Astrophysical Journal Letters, 529, L41, doi: 10.1086/312458

    work page doi:10.1086/312458 2000

  36. [36]

    2020, in Proceedings of the 37th International Conference on Machine Learning, ICML’20 (JMLR.org), doi: 10.5555/3524938.3525087

    Hinton, G., Chen, T., Kornblith, S., & Norouzi, M. 2020, in Proceedings of the 37th International Conference on Machine Learning, ICML’20 (JMLR.org), doi: 10.5555/3524938.3525087

    work page doi:10.5555/3524938.3525087 2020

  37. [37]

    2019, Astronomy & Astrophysics, 623, A39, doi: 10.1051/0004-6361/201834672

    Hippke, M., & Heller, R. 2019, Astronomy & Astrophysics, 623, A39, doi: 10.1051/0004-6361/201834672

    work page doi:10.1051/0004-6361/201834672 2019

  38. [38]

    B., Sobeck, C., Haas, M., et al

    Howell, S. B., Sobeck, C., Haas, M., et al. 2014, Publications of the Astronomical Society of the Pacific, 126, 398, doi: 10.1086/676406

    work page doi:10.1086/676406 2014

  39. [39]

    M., Caldwell, D

    Jenkins, J. M., Caldwell, D. A., & Borucki, W. J. 2002, The Astrophysical Journal, 564, 495, doi: 10.1086/324143

    work page doi:10.1086/324143 2002

  40. [40]

    M., Caldwell, D

    Jenkins, J. M., Caldwell, D. A., Chandrasekaran, H., et al. 2010b, The Astrophysical Journal Letters, 713, L87, doi: 10.1088/2041-8205/713/2/L87

    work page doi:10.1088/2041-8205/713/2/l87 2041

  41. [41]

    M., Twicken, J

    Jenkins, J. M., Twicken, J. D., Batalha, N. M., et al. 2015, The Astronomical Journal, 150, 56, doi: 10.1088/0004-6256/150/2/56

    work page doi:10.1088/0004-6256/150/2/56 2015

  42. [42]

    2020, in Proceedings of the 34th International Conference on Neural Information Processing Systems, NIPS ’20 (Red

    Khosla, P., Teterwak, P., Wang, C., et al. 2020, in Proceedings of the 34th International Conference on Neural Information Processing Systems, NIPS ’20 (Red

    2020

  43. [43]

    Hook, NY, USA: Curran Associates Inc.), doi: 10.5555/3495724.3497291

    work page doi:10.5555/3495724.3497291

  44. [44]

    Konacki, M., Torres, G., Jha, S., & Sasselov, D. D. 2003, Nature, 421, 507, doi: 10.1038/nature01379 Kov´ acs, G., Zucker, S., & Mazeh, T. 2002, Astronomy & Astrophysics, 391, 369, doi: 10.1051/0004-6361:20020802

    work page doi:10.1038/nature01379 2003

  45. [45]

    2015, Publications of the Astronomical Society of the Pacific, 127, 1161, doi: 10.1086/683602

    Kreidberg, L. 2015, Publications of the Astronomical Society of the Pacific, 127, 1161, doi: 10.1086/683602

    work page internal anchor Pith review doi:10.1086/683602 2015

  46. [46]

    S., et al

    LeCun, Y., Boser, B., Denker, J. S., et al. 1989, Neural Computation, 1, 541, doi: 10.1162/neco.1989.1.4.541

    work page doi:10.1162/neco.1989.1.4.541 1989

  47. [47]

    Laine, S

    Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. 1998, Proceedings of the IEEE, 86, 2278, doi: 10.1109/5.726791

    work page doi:10.1109/5.726791 1998

  48. [48]

    2025, Nature Astronomy, 9, 608, doi: 10.1038/s41550-025-02484-z

    Liu, T., Quan, Y., Su, Y., et al. 2025, Nature Astronomy, 9, 608, doi: 10.1038/s41550-025-02484-z

    work page doi:10.1038/s41550-025-02484-z 2025

  49. [49]

    P., & Obermeier, C

    Malik, A., Moster, B. P., & Obermeier, C. 2021, Monthly Notices of the Royal Astronomical Society, 513, 5505, doi: 10.1093/mnras/stab3692

    work page doi:10.1093/mnras/stab3692 2021

  50. [50]

    2002, The Astrophysical Journal Letters, 580, L171, doi: 10.1086/345520

    Mandel, K., & Agol, E. 2002, The Astrophysical Journal Letters, 580, L171, doi: 10.1086/345520

    work page doi:10.1086/345520 2002

  51. [51]

    1995, Nature, 378, 355, doi: 10.1038/378355a0

    Mayor, M., & Queloz, D. 1995, Nature, 378, 355, doi: 10.1038/378355a0

    work page doi:10.1038/378355a0 1995

  52. [52]

    D., Jenkins, J

    McCauliff, S. D., Jenkins, J. M., Catanzarite, J., et al. 2015, The Astrophysical Journal, 806, 6, doi: 10.1088/0004-637X/806/1/6

    work page doi:10.1088/0004-637x/806/1/6 2015

  53. [53]

    Falcon: A fourier transform based approach for fast and secure convolutional neural network predictions

    Misra, I., & van der Maaten, L. 2020, in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 6706–6716, doi: 10.1109/CVPR42600.2020.00674 NASA Exoplanet Archive. 2025, KOI Cumulative Table, IPAC, doi: 10.26133/NEA4

    work page doi:10.1109/cvpr42600.2020.00674 2020

  54. [54]

    2014, Astronomy & Astrophysics, 561, A138, doi: 10.1051/0004-6361/201220860

    Ofir, A. 2014, Astronomy & Astrophysics, 561, A138, doi: 10.1051/0004-6361/201220860

    work page doi:10.1051/0004-6361/201220860 2014

  55. [55]

    I., Rau, G., et al

    Olmschenk, G., Silva, S. I., Rau, G., et al. 2021, The Astronomical Journal, 161, 273, doi: 10.3847/1538-3881/abf4c6 DELOS in Kepler Photometry25

    work page doi:10.3847/1538-3881/abf4c6 2021

  56. [56]

    P., Ansdell, M., Ioannou, Y., et al

    Osborn, H. P., Ansdell, M., Ioannou, Y., et al. 2020, Astronomy & Astrophysics, 633, A53, doi: 10.1051/0004-6361/201935345

    work page doi:10.1051/0004-6361/201935345 2020

  57. [57]

    2021, Publications of the Astronomical Society of the Pacific, 133, 024502, doi: 10.1088/1538-3873/abd9ab

    Panahi, A., & Zucker, S. 2021, Publications of the Astronomical Society of the Pacific, 133, 024502, doi: 10.1088/1538-3873/abd9ab

    work page doi:10.1088/1538-3873/abd9ab 2021

  58. [58]

    2019, in Advances in Neural Information Processing Systems 32, 8024–8035, doi: 10.5555/3454287.3455008

    Paszke, A., Gross, S., Massa, F., et al. 2019, in Advances in Neural Information Processing Systems 32, 8024–8035, doi: 10.5555/3454287.3455008

    work page doi:10.5555/3454287.3455008 2019

  59. [59]

    A., Palafox, L., & Griffith, C

    Pearson, K. A., Palafox, L., & Griffith, C. A. 2018, Monthly Notices of the Royal Astronomical Society, 474, 478, doi: 10.1093/mnras/stx2761

    work page doi:10.1093/mnras/stx2761 2018

  60. [60]

    A., Howard, A

    Petigura, E. A., Howard, A. W., & Marcy, G. W. 2013a, Proceedings of the National Academy of Science, 110, 19273, doi: 10.1073/pnas.1319909110

    work page doi:10.1073/pnas.1319909110

  61. [61]

    A., Marcy, G

    Petigura, E. A., Marcy, G. W., & Howard, A. W. 2013b, The Astrophysical Journal, 770, 69, doi: 10.1088/0004-637X/770/1/69

    work page doi:10.1088/0004-637x/770/1/69

  62. [62]

    Gallagher, S. C. 2008, The Astrophysical Journal Supplement Series, 175, 191, doi: 10.1086/523644

    work page doi:10.1086/523644 2008

  63. [63]

    2021, Monthly Notices of the Royal Astronomical Society, 502, 2845, doi: 10.1093/mnras/stab203

    Rao, S., Mahabal, A., Rao, N., & Raghavendra, C. 2021, Monthly Notices of the Royal Astronomical Society, 502, 2845, doi: 10.1093/mnras/stab203

    work page doi:10.1093/mnras/stab203 2021

  64. [64]

    2014, Experimental Astronomy, 38, page range, doi: 10.1007/s10686-014-9383-4

    Rauer, H., Catala, C., Aerts, C., et al. 2014, Experimental Astronomy, 38, page range, doi: 10.1007/s10686-014-9383-4

    work page doi:10.1007/s10686-014-9383-4 2014

  65. [65]

    2022, in Europlanet Science Congress 2022, doi: 10.5194/epsc2022-453

    Rauer, H., Aerts, C., Deleuil, M., et al. 2022, in Europlanet Science Congress 2022, doi: 10.5194/epsc2022-453

    work page doi:10.5194/epsc2022-453 2022

  66. [66]

    2008, Astronomy & Astrophysics, 492, 617, doi: 10.1051/0004-6361:200810148

    Renner, S., Rauer, H., Erikson, A., et al. 2008, Astronomy & Astrophysics, 492, 617, doi: 10.1051/0004-6361:200810148

    work page doi:10.1051/0004-6361:200810148 2008

  67. [67]

    R., Winn, J

    Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2014, Journal of Astronomical Telescopes, Instruments, and Systems, 1, 014003, doi: 10.1117/1.JATIS.1.1.014003

    work page internal anchor Pith review doi:10.1117/1.jatis.1.1.014003 2014

  68. [68]

    C., H´ ebrard, G., et al

    Schanche, N., Cameron, A. C., H´ ebrard, G., et al. 2018, Monthly Notices of the Royal Astronomical Society, 483, 5534, doi: 10.1093/mnras/sty3146

    work page doi:10.1093/mnras/sty3146 2018

  69. [69]

    2022, Monthly Notices of the Royal Astronomical Society, 513, 2732, doi: 10.1093/mnras/stac960

    Shahaf, S., Zackay, B., Mazeh, T., Faigler, S., & Ivashtenko, O. 2022, Monthly Notices of the Royal Astronomical Society, 513, 2732, doi: 10.1093/mnras/stac960

    work page doi:10.1093/mnras/stac960 2022

  70. [70]

    J., & Vanderburg, A

    Shallue, C. J., & Vanderburg, A. 2018, The Astronomical Journal, 155, 94, doi: 10.3847/1538-3881/aa9e09

    work page doi:10.3847/1538-3881/aa9e09 2018

  71. [71]

    2020, arXiv e-prints, arXiv:2011.10944, doi: 10.48550/arXiv.2011.10944

    Shi, H., Luo, D., Tang, S., Wang, J., & Zhuang, Y. 2020, arXiv e-prints, arXiv:2011.10944, doi: 10.48550/arXiv.2011.10944

    work page doi:10.48550/arxiv.2011.10944 2020

  72. [72]

    C., Stumpe, M

    Smith, J. C., Stumpe, M. C., Cleve, J. E. V., et al. 2012, Publications of the Astronomical Society of the Pacific, 124, 1000, doi: 10.1086/667697

    work page internal anchor Pith review doi:10.1086/667697 2012

  73. [73]

    1969, Proceedings of the IEEE, 57, 724, doi: 10.1109/PROC.1969.7051

    Staelin, D. 1969, Proceedings of the IEEE, 57, 724, doi: 10.1109/PROC.1969.7051

    work page doi:10.1109/proc.1969.7051 1969

  74. [74]

    C., Smith, J

    Stumpe, M. C., Smith, J. C., Cleve, J. E. V., et al. 2012, Publications of the Astronomical Society of the Pacific, 124, 985, doi: 10.1086/667698

    work page internal anchor Pith review doi:10.1086/667698 2012

  75. [75]

    E., Coughlin, J

    Thompson, S. E., Coughlin, J. L., Hoffman, K., et al. 2018, The Astrophysical Journal Supplement Series, 235, 38, doi: 10.3847/1538-4365/aab4f9

    work page doi:10.3847/1538-4365/aab4f9 2018

  76. [76]

    VanderPlas, J. T. 2018, The Astrophysical Journal Supplement Series, 236, doi: 10.3847/1538-4365/aab766 von Braun, K., & Ciardi, D. R. 2007, Proceedings of the International Astronomical Union, 3, 93–100, doi: 10.1017/S1743921308016451

    work page internal anchor Pith review doi:10.3847/1538-4365/aab766 2018

  77. [77]

    2024a, Monthly Notices of the Royal Astronomical Society, 528, 4053, doi: 10.1093/mnras/stae245

    Wang, K., Ge, J., Willis, K., Wang, K., & Zhao, Y. 2024a, Monthly Notices of the Royal Astronomical Society, 528, 4053, doi: 10.1093/mnras/stae245

    work page doi:10.1093/mnras/stae245

  78. [78]

    2024b, Monthly Notices of the Royal Astronomical Society, 534, 1913, doi: 10.1093/mnras/stae2155

    Wang, K., Ge, J., Willis, K., et al. 2024b, Monthly Notices of the Royal Astronomical Society, 534, 1913, doi: 10.1093/mnras/stae2155

    work page doi:10.1093/mnras/stae2155 1913

  79. [79]

    2019, The Astronomical Journal, 158, 25, doi: 10.3847/1538-3881/ab21d6

    Yu, L., Vanderburg, A., Huang, C., et al. 2019, The Astronomical Journal, 158, 25, doi: 10.3847/1538-3881/ab21d6

    work page doi:10.3847/1538-3881/ab21d6 2019

  80. [80]

    2018, The Astronomical Journal, 155, 147, doi: 10.3847/1538-3881/aaae05

    Zucker, S., & Giryes, R. 2018, The Astronomical Journal, 155, 147, doi: 10.3847/1538-3881/aaae05

    work page doi:10.3847/1538-3881/aaae05 2018