Pith. sign in

REVIEW 2 major objections 5 minor 300 references

A multi-band Faster R-CNN finds giant star-forming clumps in nearby galaxies at completeness ≳ 0.9 and purity ≳ 0.8.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-11 21:09 UTC pith:VTB2RB6K

load-bearing objection Solid multi-band FRCNN + Zoobot detector with clean completeness numbers and a large public catalogue; purity is less independent than advertised but still usable. the 2 major comments →

arxiv 2607.04176 v1 pith:VTB2RB6K submitted 2026-07-05 astro-ph.IM astro-ph.GA

Star-forming clump detection in nearby galaxies using Faster R-CNN and ugrizy imaging data from CLAUDS and HSC-SSP

J\"urgen J. Popp , Hugh Dickinson , Stephen Serjeant , Lucy F. Fortson , Tobias G\'eron , Brooke D. Simmons , Vihang Mehta This is my paper
classification astro-ph.IM astro-ph.GA
keywords star-forming clumpsobject detectionFaster R-CNNtransfer learninggalaxy morphologyCLAUDSHSC-SSPdeep learning
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Giant star-forming clumps are common at high redshift but hard to find in large numbers of nearby galaxies. This paper shows that a Faster R-CNN object detector, expanded to take six simultaneous ugrizy images and built on a pretrained Zoobot backbone, can locate those clumps and distinguish them from stars, bulges and other contaminants. On real CLAUDS and HSC-SSP galaxies into which realistic simulated clumps were injected, the detector recovers more than 90 percent of clumps brighter than the survey depth limits while keeping purity above 80 percent. The same model then yields roughly 1.5 million clump candidates across hundreds of thousands of low-redshift galaxies, opening a statistical sample of local analogues that had previously been limited to a few hundred objects.

Core claim

An expanded Faster R-CNN that simultaneously processes six ugrizy filter-band images and classifies six object types, using Zoobot as its feature-extraction backbone, detects simulated star-forming clumps brighter than the 5σ survey limits with completeness ≳ 0.9 and purity ≳ 0.8, thereby enabling a catalogue of ~1.5 million clump candidates in ~347 000 low-redshift galaxies.

What carries the argument

The multi-channel Faster R-CNN (FRCNN) with Zoobot ResNet50 backbone: the first convolutional block is widened to accept five or six filter bands, the detector head predicts six classes (clump, odd clump, star, background galaxy, bulge, background), and detections are refined by non-maximum suppression, size cuts and flux-peak extraction inside each bounding box.

Load-bearing premise

That a few thousand human-corrected training labels, plus clumps simulated with stellar-population models and injected into real images, faithfully represent the true low-redshift clump population so that the measured completeness and purity transfer to real detections.

What would settle it

A carefully inspected subsample of several hundred galaxies at z ≲ 0.05 with independent high-resolution imaging or spectroscopy that shows the fraction of true giant clumps recovered (or the fraction of contaminants) falls well below the claimed 0.9/0.8 thresholds.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Statistical studies of clump mass, colour and radial distribution become possible for the first time at z ≲ 0.5 with sample sizes of order 10^6.
  • The same multi-channel architecture can be applied directly to the full HSC-SSP Wide survey area without requiring new u-band data for every galaxy.
  • Future higher-resolution surveys can retrain the same backbone with only modest additional labelling and immediately improve recovery of faint or sub-clump structure.
  • Photometric and physical-property catalogues released with the detections provide ready-made inputs for follow-up SED fitting and comparison with high-redshift clump samples.

Where Pith is reading between the lines

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

  • Because completeness remains high even when the u-band is dropped, the method can be ported to any deep optical multi-band survey that lacks a dedicated NUV channel.
  • The residual false positives that cluster near galaxy centres suggest a natural next filter: a simple radial or contrast cut that would raise purity without retraining.
  • If the same architecture is fine-tuned on Euclid or Roman imaging, the resulting catalogue will bridge the resolution gap between local and high-redshift clump studies.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 5 minor

Summary. The paper presents a multi-channel Faster R-CNN object detector that uses a Zoobot (ResNet50) backbone to locate and classify giant star-forming clumps (and several contaminant classes) in CLAUDS+HSC-SSP ugrizy imaging of low-redshift (z≤0.5) galaxies. Two models (5- and 6-channel) are trained on ~3 200 human-corrected labels, post-processed with NMS, a 7.3 kpc size cut and a galaxy segmentation mask, and then applied to ~347 000 galaxies, yielding a catalogue of ~1.5 million clump candidates. Completeness and purity are quantified by injecting ~32 000 FSPS-simulated clumps into ~14 000 real galaxies; for objects brighter than the survey 5σ depths the authors report completeness ≳0.9 and purity ≳0.8.

Significance. If the performance numbers hold, the work supplies one of the first large, publicly usable catalogues of low-redshift GSFCs and demonstrates a practical downstream use of the Zoobot foundation model for multi-band object detection. The multi-channel architecture, multi-class contaminant scheme and careful injection tests are genuine methodological advances over earlier binary or single-band approaches. The accompanying code and training data further raise the paper’s value for the community.

major comments (2)
  1. §6.3.3–6.3.4 and Tables 6–7: purity is obtained by first masking every pre-injection detection and then counting residual detections that match the injected clumps. The residual set can still contain real clumps the original model missed plus injection-triggered false positives, so the quoted purity (≳0.8 above the 5σ limit) is an upper bound conditioned on the same detector. Completeness is cleanly measured; purity is not. A clearer statement of this limitation (or an independent purity estimate, e.g. from a fully held-out visual sample) is needed before the catalogue contamination floor can be trusted.
  2. §3.2 and §7: the training set of only ~3 200 galaxies (few at z≲0.05) was built by correcting an earlier machine prototype. The authors themselves note that faint, well-resolved clumps may have been missed. Because both the detector and the purity denominator inherit this bias, the claim that the measured completeness/purity transfer to the full ~1.5 M-candidate catalogue rests on an incompletely tested assumption about the representativeness of the labels and the FSPS injections.
minor comments (5)
  1. Figure 1 and surrounding text: the over-fitting discussion would be clearer if the epoch of the final checkpoint (and the corresponding F1) were marked on every panel.
  2. Equation (1) and the Lupton scaling (Eq. 2) are adopted from Zoobot pre-training; a short justification that these scalings remain optimal for multi-band clump detection would help.
  3. Table 2: the parenthetical “clump candidates per galaxy” ratios would be easier to interpret if the number of clumpy galaxies (rather than total galaxies) were also listed.
  4. Appendix C visual inspection of 579 galaxies is useful; stating the redshift and mass distribution of that subsample would strengthen the false-positive/false-negative fractions quoted there.
  5. A few typographical slips remain (e.g. “the the Faster”, “greater then”, inconsistent use of “GSFC” vs “clump”).

Circularity Check

1 steps flagged

No load-bearing circular derivation; empirical completeness/purity on independent FSPS injections is self-contained, with only mild self-reference in the human-corrected training labels.

specific steps
  1. other [Section 3.2 (Training data)]
    "we overlaid the galaxy images with a first set of possible clump detections. This set of possible clump detections were the predictions from a model with the same architecture as the final 6-channel model … By adopting a limited ‘correct-a-machine’ approach, we expected that the final detection model will benefit from being trained not only on new clump labels but also on its previous errors."

    The initial training annotations begin from detections of a prototype of identical architecture; human corrections therefore start from a distribution already shaped by the same model family. This is a mild self-reference in the training distribution, not a definitional or fitted-input circularity that forces the later completeness/purity numbers (which are measured on independent injections).

full rationale

The paper is an empirical methods paper whose central claims (completeness ≳0.9 and purity ≳0.8 for clumps brighter than the survey 5σ depths) are measured by injecting FSPS-simulated clumps whose physical parameters (stellar mass, age, AV, metallicity, SFH τ) are drawn from independent distributions (Table 3) into real galaxy images after deliberately masking the model’s own pre-injection detections, then recovering flux peaks (Sections 6.2–6.3, Tables 4–7). Completeness is simply the fraction of known injected positions recovered within 0.75× seeing FWHM; purity is the fraction of residual post-masking detections that match those injections. Neither quantity is forced by construction or by a fitted parameter renamed as a prediction. Training labels are human-corrected annotations on ~3200 galaxies (Section 3.2), not pure self-labels, and the Zoobot backbone is an external foundation model. The only mild self-reference is the limited “correct-a-machine” overlay of a prototype of the same architecture, which the authors themselves flag as a possible bias source; this does not reduce the reported metrics to their inputs. No uniqueness theorems, ansatz smuggling, or renaming of known results appear. Score 1 reflects that single non-load-bearing self-reference; the derivation chain itself is free of circularity.

Axiom & Free-Parameter Ledger

5 free parameters · 4 axioms · 0 invented entities

The central performance claim rests on standard cosmological and stellar-population assumptions, a set of hand-chosen detection and simulation thresholds, and the premise that human-corrected labels plus FSPS injections are representative. No new physical entities are postulated; free parameters are algorithmic cut-offs and sampling ranges rather than fitted physical constants.

free parameters (5)
  • Maximum bounding-box size cut = 7.30 kpc
    Set to 7.30 kpc (95th percentile of the size distribution after visual inspection of a few hundred galaxies); used to discard oversized detections (Section 4).
  • NMS IoU threshold = 0.2
    IoU ≥ 0.2 used for non-maximum suppression of overlapping boxes (Section 4).
  • Matching radius for simulated clumps = 0.75 × FWHM
    0.75 × image-specific u-band seeing FWHM chosen from the distance distribution (Section 6.3.1).
  • Simulated clump stellar-mass upper limits by redshift bin = redshift-dependent caps
    Log-uniform sampling capped at 10^7, 5 imes10^7, 10^8, 5 imes10^8 M☉ for successive redshift bins and at 0.1 M_galaxy (Table 3).
  • Objectness score thresholds for performance tables = 0.0 / 0.3 / 0.6
    Reported at cn ≥ 0.0, 0.3, 0.6; final catalogue uses no hard cut but relies on these for model selection (Table 1).
axioms (4)
  • domain assumption Planck 2015 cosmology (Ωm, ΩΛ, h) = (0.31, 0.69, 0.68) for converting angular sizes to physical kpc.
    Stated in the introduction and used for the 7.30 kpc size cut and radial-distance plots.
  • domain assumption FSPS delayed-τ CSPs with Calzetti dust, Chabrier IMF, MIST/MILES libraries adequately reproduce the broadband colours and luminosities of real GSFCs.
    Section 6.2; completeness and purity are measured exclusively against these injections.
  • domain assumption Human visual labels (including corrections of a prototype model) define the ground-truth class of clumps versus contaminants.
    Section 3.2; the entire supervised training objective is defined by these labels.
  • ad hoc to paper asinh and Lupton scalings used for Zoobot pre-training remain optimal for multi-band clump detection.
    Section 2.2; chosen to match Zoobot's training images rather than re-optimised for detection.

pith-pipeline@v1.1.0-grok45 · 40926 in / 3310 out tokens · 29731 ms · 2026-07-11T21:09:26.667050+00:00 · methodology

0 comments
read the original abstract

Giant Star-forming Clumps (GSFCs) are kpc-scale regions of enhanced star-formation with stellar masses of $10^7$ to $10^9\,M_\odot$ that are commonly observed in high-redshift galaxies but are rarely detected in low-redshift ($z\lesssim0.5$) galaxy analogues. However, the availability of wide-field galaxy survey data makes it possible to identify potential star-forming clumps in large samples of low-redshift galaxies using object detection models that are based on Deep Learning (DL) techniques. We apply a novel DL-based object detection model to galaxies observed by the Hyper Suprime-Cam Subaru Strategic Survey (HSC-SSP) and CFHT Large Area U-band Deep Survey (CLAUDS). Our model is based on the the Faster Region-Based Convolutional Neural Network (Faster R-CNN or FRCNN) object detection framework but expanded to process the six $ugrizy$ filter band images simultaneously and identify not only clumps and their locations in the host galaxy but also additional contaminants. By adopting the \textsc{Zoobot} foundation DL-model as a feature extraction backbone, we also demonstrate one of the first applications of \textsc{Zoobot} in a downstream task for object detection. Our model achieves a detection completeness of $\gtrsim 0.9$ and purity of $\gtrsim 0.8$ which were validated on a large set of real galaxies into which simulated clumps were injected.

Figures

Figures reproduced from arXiv: 2607.04176 by Brooke D. Simmons, Hugh Dickinson, J\"urgen J. Popp, Lucy F. Fortson, Stephen Serjeant, Tobias G\'eron, Vihang Mehta.

Figure 1
Figure 1. Figure 1: Training and validation losses of the FRCNN models for different training modes. The training and validation losses for the 6-channel model are shown in the plots at the top with the model trained in fine-tuning mode on the left and in transfer mode on the right. The bottom plots show the losses of the 5-channel model for the same training modes. The loss determined on the validation data is plotted in ora… view at source ↗
Figure 2
Figure 2. Figure 2: Plot showing the 𝐹1 scores at each epoch for the 6-channel model that are trained in fine-tuning mode (blue) and in transfer learning mode (orange) as well as for the 5-channel models that are trained in fine-tuning mode (grey) and in transfer learning mode (black). The epochs with the highest 𝐹1 score before the models start to over-fit are indicated by dotted vertical lines in the same colours. training … view at source ↗
Figure 3
Figure 3. Figure 3: Precision and recall of the FRCNN models for different objectness thresholds. The objectness threshold 𝑐𝑛 is increasing from 0.0 (right) to 0.99 (left) as indicated by the annotations. Error bars show the 95% confidence interval. The red square in panel (a) marks the zoomed-in area shown in panel (b). of IoU ≥ 0.2 and kept only those detections and the corresponding object class predictions that have the h… view at source ↗
Figure 4
Figure 4. Figure 4: Example galaxy (object 43153778395866061, z=0.03) showing the different postprocessing steps applied to the FRCNN model detections. The left image shows the raw detection results from the model, the central image the detection results after the non-maximum suppression process was applied and the right image after bounding boxes with size > 7.30 kpc were removed. host galaxy. Although, the training data was… view at source ↗
Figure 5
Figure 5. Figure 5: Three galaxy examples showing the postprocessed FRCNN model detections with extracted flux peaks. The model detections are shown as boxes where the colour indicates the object class. Flux peaks are marked with red crosses. The galaxies are shown with their u-band images [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of the clump detections from both models and detec￾tions unique to either the ugrizy or grizy model. The kernel density estimates of the clump detections are shown separately as a function of the objectness score that is output by the 6-channel (a) and 5-channel model (b). Clump can￾didates that were detected by the 6-channel model and the 5-channel model are shown in blue (a) and in grey (b). C… view at source ↗
Figure 7
Figure 7. Figure 7: Detections from the 5- and 6-channel FRCNN models shown for an example of ten galaxies. The galaxies are shown in pairs with the u-band image first and followed by the g-band image of the same galaxy. Detections from the 6-channel FRCNN model are marked with orange circles and the detections from the 5-channel model with green circles. Non-clump detections (e.g. fore-/background galaxies, stars) are marked… view at source ↗
Figure 8
Figure 8. Figure 8: Redshift vs. i-band seeing FWHM for HSC-SSP galaxies and their image cutouts. The red line plots the theoretical seeing FWHM that is required to resolve objects with 1.0 kpc in physical size. The marginal histograms on the top and right side of the two-dimensional histogram show the univariate distributions for all galaxies in blue and galaxies for which the i-band seeing FWHM translates to a physical size… view at source ↗
Figure 9
Figure 9. Figure 9: Stellar mass vs. specific star-formation rate (sSFR) of the host galaxies with bounding box detections that contain different numbers of identified clumps (rows) and per redshift bin (columns). The distribution of the host galaxies are shown in coloured hex-bins. The horizontal dashed line marks the separation between star-forming and quiescent galaxies at log(sSFR/yr−1 ) = −11.0 (e.g. McGee et al. 2011; W… view at source ↗
Figure 10
Figure 10. Figure 10: Histograms showing the distribution of the sampled parameters for the final set of simulated clumps. The clump stellar masses (a) were sampled from a log-uniform distribution with a lower limit of 104 𝑀⊙ and upper limit of 0.1 𝑀galaxy (see also [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Galaxy examples with simulated clumps, shown as the original u-band science image (first column), the simulated clumps alone (second column), u-band science image with simulated clumps injected (third column) and the RGB-composite image (generated using the GRI-bands) with injected simulated clumps (last column). The positions of the simulated clumps are indicated by small red markers in the images of the… view at source ↗
Figure 12
Figure 12. Figure 12: RGB-composite images (generated using the GRI-bands) showing the model detections with simulated clumps injected. The positions of the simulated clumps are indicated by small red markers in the images and the model detections are shown as bounding boxes around clump complexes (solid boxes: detections on the real images, dotted boxes: additional detections from the images with injected clumps). Some detect… view at source ↗
Figure 13
Figure 13. Figure 13: Distribution of the distance in arcsec between the simulated clumps and the closest detections. The distribution of those clumps that are considered a matching or successful detection are shown in orange. 20 22 24 26 28 30 Simulated clump magnitude [mAB] 0.0 0.2 0.4 0.6 0.8 1.0 Completeness u g r i z y (a) Model completeness per 𝑚AB-bin of the 6-channel (ugrizy) FRCNN model. 20 22 24 26 28 30 Simulated cl… view at source ↗
Figure 15
Figure 15. Figure 15: Detection completeness of the simulated clumps as a function of different physical clump properties for the 6-channel FRCNN model. The completeness is plotted in blue for the full sample of simulated clumps and in orange for clumps that are brighter than the detection limit of the two surveys (𝑢 ≤ 27.1 𝑚AB, 𝑔 ≤ 26.5 𝑚AB, 𝑟 ≤ 26.1 𝑚AB, 𝑖 ≤ 25.9 𝑚AB, 𝑧 ≤ 25.1 𝑚AB and 𝑦 ≤ 24.4 𝑚AB). The shaded areas show the… view at source ↗
Figure 14
Figure 14. Figure 14: Detection completeness of the 6-channel (a) and 5-channel (b) FRCNN model with respect to the simulated clumps. The plots show the fraction of detected simulated clumps in magnitude 𝑚AB-bins with bin widths of 0.2 𝑚AB. Shaded areas show the 95% confidence interval and the dotted vertical lines mark the detection limits for each filter band. (𝑚AB ≲ 21), the estimated completeness varies more due to the low… view at source ↗
Figure 16
Figure 16. Figure 16: Similar to [PITH_FULL_IMAGE:figures/full_fig_p016_16.png] view at source ↗
Figure 19
Figure 19. Figure 19: Purity of the 6-channel (a) and 5-channel (b) FRCNN model measured on simulated clumps as a function of apparent magnitude. The plots show the fraction of detected simulated clumps to all detections per magnitude 𝑚AB-bins with bin widths of 0.2 𝑚AB. Shaded areas show the 95% confidence interval and the dotted vertical lines mark the detection limits for each filter band. not representative of the true det… view at source ↗
Figure 18
Figure 18. Figure 18: Similar to [PITH_FULL_IMAGE:figures/full_fig_p017_18.png] view at source ↗
Figure 20
Figure 20. Figure 20: Purity as a function of different physical clump and host galaxy properties measured on simulated clumps for the 6-channel FRCNN model. The purity is plotted in blue for the full sample of detected clumps and in orange for clumps that are brighter than the detection limit of the two surveys (𝑢 ≤ 27.1 𝑚AB, 𝑔 ≤ 26.5 𝑚AB, 𝑟 ≤ 26.1 𝑚AB, 𝑖 ≤ 25.9 𝑚AB, 𝑧 ≤ 25.1 𝑚AB and 𝑦 ≤ 24.4 𝑚AB). The shaded areas show the 1… view at source ↗
Figure 21
Figure 21. Figure 21: Purity as a function of different physical clump and host galaxy properties measured on simulated clumps for the 5-channel FRCNN model. The purity is plotted in blue for the full sample of detected clumps and in orange for clumps that are brighter than the detection limit for the HSC-SSP filter bands (𝑔 ≤ 26.5 𝑚AB, 𝑟 ≤ 26.1 𝑚AB, 𝑖 ≤ 25.9 𝑚AB, 𝑧 ≤ 25.1 𝑚AB and 𝑦 ≤ 24.4 𝑚AB). The shaded areas show the 1𝜎 er… view at source ↗

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

300 extracted references · 58 canonical work pages · 22 internal anchors

  1. [1]

    2022 , month = may, number =

    Adams, Dominic and Mehta, Vihang and Dickinson, Hugh and Scarlata, Claudia and Fortson, Lucy and Kruk, Sandor and Simmons, Brooke and Lintott, Chris , journal =. 2022 , month = may, number =. doi:10.3847/1538-4357/ac6512 , eid =

    doi:10.3847/1538-4357/ac6512 2022

  2. [2]

    2010 , month = may, doi =

    Houjun Mo and Frank van den Bosch and Simon White , publisher =. 2010 , month = may, doi =

    2010

  3. [3]

    Elmegreen and Clara M

    Debra Meloy Elmegreen and Bruce G. Elmegreen and Clara M. Sheets , journal =. 2004 , month = mar, number =. doi:10.1086/381357 , file =

    work page doi:10.1086/381357 2004

  4. [4]

    Elmegreen and Debra Meloy Elmegreen , journal =

    Bruce G. Elmegreen and Debra Meloy Elmegreen , journal =. 2005 , month = jul, number =. doi:10.1086/430514 , file =

    doi:10.1086/430514 2005

  5. [5]

    Elmegreen and Fr

    Bruce G. Elmegreen and Fr. The Astrophysical Journal , title =. 2008 , month = nov, number =. doi:10.1086/592190 , file =

    doi:10.1086/592190 2008

  6. [6]

    Ferguson and Eric F

    Yicheng Guo and Henry C. Ferguson and Eric F. Bell and David C. Koo and Christopher J. Conselice and Mauro Giavalisco and Susan Kassin and Yu Lu and Ray Lucas and Nir Mandelker and Daniel M. McIntosh and Joel R. Primack and Swara Ravindranath and Guillermo Barro and Daniel Ceverino and Avishai Dekel and Sandra M. Faber and Jerome J. Fang and Anton M. Koek...

    doi:10.1088/0004-637x/800/1/39 2015

  7. [7]

    Bell and Christopher J

    Yicheng Guo and Marc Rafelski and Eric F. Bell and Christopher J. Conselice and Avishai Dekel and S. M. Faber and Mauro Giavalisco and Anton M. Koekemoer and David C. Koo and Yu Lu and Nir Mandelker and Joel R. Primack and Daniel Ceverino and Duilia F. de Mello and Henry C. Ferguson and Nimish Hathi and Dale Kocevski and Ray A. Lucas and Pablo G. P. The A...

    doi:10.3847/1538-4357/aaa018 2018

  8. [8]

    de Mello and Marc Rafelski and Jonathan P

    Emmaris Soto and Duilia F. de Mello and Marc Rafelski and Jonathan P. Gardner and Harry I. Teplitz and Anton M. Koekemoer and Swara Ravindranath and Norman A. Grogin and Claudia Scarlata and Peter Kurczynski and Eric Gawiser , journal =. 2017 , month = feb, number =. doi:10.3847/1538-4357/aa5da3 , file =

    doi:10.3847/1538-4357/aa5da3 2017

  9. [9]

    2018 , month = jun, number =

    Miroslava Dessauges-Zavadsky and Angela Adamo , journal =. 2018 , month = jun, number =. doi:10.1093/mnrasl/sly112 , file =

    doi:10.1093/mnrasl/sly112 2018

  10. [10]

    Fisher and Karl Glazebrook and Ivana Damjanov and Roberto G

    David B. Fisher and Karl Glazebrook and Ivana Damjanov and Roberto G. Abraham and Danail Obreschkow and Emily Wisnioski and Robert Bassett and Andy Green and Peter McGregor , journal =. 2017 , month = sep, number =. doi:10.1093/mnras/stw2281 , file =

    doi:10.1093/mnras/stw2281 2017

  11. [11]

    The Astrophysical Journal , title =

    Vihang Mehta and Claudia Scarlata and Lucy Fortson and Hugh Dickinson and Dominic Adams and Jacopo Chevallard and St. The Astrophysical Journal , title =. 2021 , month = may, number =. doi:10.3847/1538-4357/abed5b , file =

    doi:10.3847/1538-4357/abed5b 2021

  12. [12]

    High-redshift clumpy discs and bulges in cosmological simulations , year =

    Daniel Ceverino and Avishai Dekel and Frederic Bournaud , journal =. High-redshift clumpy discs and bulges in cosmological simulations , year =. doi:10.1111/j.1365-2966.2010.16433.x , groups =

    doi:10.1111/j.1365-2966.2010.16433.x 2010

  13. [13]

    Rotational support of giant clumps in high-z disc galaxies , year =

    Daniel Ceverino and Avishai Dekel and Nir Mandelker and Frederic Bournaud and Andreas Burkert and Reinhard Genzel and Joel Primack , journal =. Rotational support of giant clumps in high-z disc galaxies , year =. doi:10.1111/j.1365-2966.2011.20296.x , groups =

    doi:10.1111/j.1365-2966.2011.20296.x 2011

  14. [14]

    Tobias Buck and Andrea V. Macci. Monthly Notices of the Royal Astronomical Society , title =. 2017 , month = mar, number =. doi:10.1093/mnras/stx685 , file =

    doi:10.1093/mnras/stx685 2017

  15. [15]

    Gnedin , journal =

    Xi Meng and Oleg Y. Gnedin , journal =. Origin of giant stellar clumps in high-redshift galaxies , year =. doi:10.1093/mnras/staa776 , groups =

    doi:10.1093/mnras/staa776

  16. [16]

    Romeo and Oscar Agertz , journal =

    Florent Renaud and Alessandro B. Romeo and Oscar Agertz , journal =. 2021 , month = sep, number =. doi:10.1093/mnras/stab2604 , groups =

    doi:10.1093/mnras/stab2604 2021

  17. [17]

    From giant clumps to clouds -

    Floor van Donkelaar and Oscar Agertz and Florent Renaud , journal =. From giant clumps to clouds -. 2022 , month = mar, number =. doi:10.1093/mnras/stac692 , groups =

    doi:10.1093/mnras/stac692 2022

  18. [18]

    R. C. Livermore and T. Jones and J. Richard and R. G. Bower and R. S. Ellis and A. M. Swinbank and J. R. Rigby and Ian Smail and S. Arribas and J. Rodriguez-Zaurin and L. Colina and H. Ebeling and R. A. Crain , journal =. 2012 , month = nov, number =. doi:10.1111/j.1365-2966.2012.21900.x , file =

    doi:10.1111/j.1365-2966.2012.21900.x 2012

  19. [19]

    2017 , month = feb, number =

    Miroslava Dessauges-Zavadsky and Daniel Schaerer and Antonio Cava and Lucio Mayer and Valentina Tamburello , journal =. 2017 , month = feb, number =. doi:10.3847/2041-8213/aa5d52 , file =

    doi:10.3847/2041-8213/aa5d52 2017

  20. [20]

    R. J. Ivison and J. Richard and A. D. Biggs and M. A. Zwaan and E. Falgarone and V. Arumugam and P. P. van der Werf and W. Rujopakarn , journal =. 2020 , month = mar, number =. doi:10.1093/mnrasl/slaa046 , groups =

    doi:10.1093/mnrasl/slaa046 2020

  21. [21]

    Mike Walmsley and Anna M. M. Scaife and Chris Lintott and Michelle Lochner and Verlon Etsebeth and Tobias G. Monthly Notices of the Royal Astronomical Society , title =. 2022 , month = feb, number =. doi:10.1093/mnras/stac525 , file =

    work page doi:10.1093/mnras/stac525 2022

  22. [22]

    Monthly Notices of the Royal Astronomical Society , title =

    Mike Walmsley and Chris Lintott and Tobias G. Monthly Notices of the Royal Astronomical Society , title =. 2021 , month = sep, number =. doi:10.1093/mnras/stab2093 , file =

    doi:10.1093/mnras/stab2093 2021

  23. [23]

    2019 , groups =

    Shilpa Ananth , month = aug, title =. 2019 , groups =

    2019

  24. [24]

    2007 , editor =

    Elmegreen, Debra Meloy , booktitle =. 2007 , editor =. doi:10.1017/S1743921306010210 , file =

    work page doi:10.1017/s1743921306010210 2007

  25. [25]

    2020 , month = jan, number =

    Walmsley, Mike and Smith, Lewis and Lintott, Chris and Gal, Yarin and Bamford, Steven and Dickinson, Hugh and Fortson, Lucy and Kruk, Sandor and Masters, Karen and Scarlata, Claudia and Simmons, Brooke and Smethurst, Rebecca and Wright, Darryl , journal =. 2020 , month = jan, number =. arXiv , doi =:1905.07424 , groups =

    Pith/arXiv arXiv 2020

  26. [26]

    2022 , month = nov, abstract =

    Shoubaneh Hemmati and Eric Huff and Hooshang Nayyeri and Agnès Ferté and Peter Melchior and Bahram Mobasher and Jason Rhodes and Abtin Shahidi and Harry Teplitz , title =. 2022 , month = nov, abstract =

    2022

  27. [27]

    Zoobot: Deep learning galaxy morphology classifier , year =

    Walmsley, Mike and Lintott, Chris and G. Zoobot: Deep learning galaxy morphology classifier , year =. ascl , eid =:2203.027 , groups =

  28. [28]

    and Vaccari, Mattia , booktitle =

    Fielding, Ezra and Nyirenda, Clement N. and Vaccari, Mattia , booktitle =. A Comparison of Deep Learning Architectures for Optical Galaxy Morphology Classification , year =. arXiv , doi =:2111.04353 , groups =

    Pith/arXiv arXiv

  29. [29]

    2014 , month = aug, pages =

    Madau, Piero and Dickinson, Mark , journal =. 2014 , month = aug, pages =. doi:10.1146/annurev-astro-081811-125615 , file =

    doi:10.1146/annurev-astro-081811-125615 2014

  30. [30]

    2022 , month = nov, abstract =

    Heng Yu and Xiaolan Hou , title =. 2022 , month = nov, abstract =. arXiv , doi =:2211.06002 , file =

    Pith/arXiv arXiv 2022

  31. [31]

    Flaccomio and G

    E. Flaccomio and G. Micela and G. Peres and S. Sciortino and E. Salvaggio and L. Prisinzano and M. G. Guarcello and L. Venuti and R. Bonito and I. Pillitteri , title =. 2022 , month = nov, abstract =

    2022

  32. [32]

    2022 , month = nov, abstract =

    Fabrizio Fiore and Andrea Ferrara and Manuela Bischetti and Chiara Feruglio and Andrea Travascio , title =. 2022 , month = nov, abstract =

    2022

  33. [33]

    Sorce and Antonino Troja , title =

    Isaac Tutusaus and Jenny G. Sorce and Antonino Troja , title =. 2022 , month = nov, abstract =

    2022

  34. [34]

    2022 , month = nov, abstract =

    Suraj Dhiwar and Kanak Saha and Avishai Dekel and Abhishek Paswan and Divya Pandey and Arianna Cortesi and Mahadev Pandge , title =. 2022 , month = nov, abstract =

    2022

  35. [35]

    J. L. Tous and H. Domínguez-Sánchez and J. M. Solanes and J. D. Perea , title =. 2022 , month = nov, abstract =

    2022

  36. [36]

    Sorce , title =

    Antonino Troja and Isaac Tutusaus and Jenny G. Sorce , title =. 2022 , month = nov, abstract =

    2022

  37. [37]

    2022 , month = nov, abstract =

    Utsav Akhaury and Jean-Luc Starck and Pascale Jablonka and Frédéric Courbin and Kevin Michalewicz , title =. 2022 , month = nov, abstract =. arXiv , doi =:2211.09597 , file =

    Pith/arXiv arXiv 2022

  38. [38]

    B. D. Simmons and Chris Lintott and Kyle W. Willett and Karen L. Masters and Jeyhan S. Kartaltepe and Boris Häu. Monthly Notices of the Royal Astronomical Society , title =. 2016 , month = oct, number =. doi:10.1093/mnras/stw2587 , groups =

    doi:10.1093/mnras/stw2587 2016

  39. [39]

    Lean Crowdsourcing: Combining Humans and Machines in an Online System , year =

    Steve Branson and Grant Van Horn and Pietro Perona , booktitle =. Lean Crowdsourcing: Combining Humans and Machines in an Online System , year =. doi:10.1109/cvpr.2017.647 , groups =

    work page doi:10.1109/cvpr.2017.647 2017

  40. [40]

    Sorce and Antonino Troja and Isaac Tutusaus , title =

    Jenny G. Sorce and Antonino Troja and Isaac Tutusaus , title =. 2022 , month = nov, abstract =

    2022

  41. [41]

    2022 , month = nov, abstract =

    Michele Delli Veneri and Lukasz Tychoniec and Fabrizia Guglielmetti and Giuseppe Longo and Eric Villard , title =. 2022 , month = nov, abstract =. arXiv , doi =:2211.11462 , file =

    Pith/arXiv arXiv 2022

  42. [42]

    Monthly Notices of the Royal Astronomical Society , title =

    Joshua Wilde and Stephen Serjeant and Jane M Bromley and Hugh Dickinson and L. Monthly Notices of the Royal Astronomical Society , title =. 2022 , month = feb, number =. doi:10.1093/mnras/stac562 , groups =

    doi:10.1093/mnras/stac562 2022

  43. [43]

    AGNet: Weighing Black Holes with Deep Learning , year =

    Joshua Yao-Yu Lin and Sneh Pandya and Devanshi Pratap and Xin Liu and Matias Carrasco Kind and Volodymyr Kindratenko , journal =. AGNet: Weighing Black Holes with Deep Learning , year =. arXiv , doi =:2108.07749 , file =

    Pith/arXiv arXiv

  44. [44]

    Poggianti and the GASP team , title =

    Bianca M. Poggianti and the GASP team , title =. 2022 , month = nov, abstract =

    2022

  45. [45]

    2015 , editor =

    Ren, Shaoqing and He, Kaiming and Girshick, Ross and Sun, Jian , booktitle =. 2015 , editor =

    2015

  46. [46]

    A chemical study of nine star-forming regions with evidence of infall motion , year =

    Yang Yang and Yao Wang and Zhibo Jiang and Zhiwei Chen , journal =. A chemical study of nine star-forming regions with evidence of infall motion , year =. doi:10.1093/mnras/stac3130 , groups =

    work page doi:10.1093/mnras/stac3130

  47. [47]

    Jimena Rodriguez and Janice Lee and Bradley Whitmore and David Thilker and Daniel Maschmann and Rupali Chandar and Daniel Dale and Diederik Kruijssen and Mederic Boquien and Kathryn Grasha and Elizabeth Watkins and Ashley Barnes and Mattia Sormani and Thomas Williams and Jaeyeon Kim and Gagandeep Anand and Mélanie Chevance and Frank Bigiel and Adam Leroy ...

    2022

  48. [48]

    and Lintott, Chris J

    Willett, Kyle W. and Lintott, Chris J. and Bamford, Steven P. and Masters, Karen L. and Simmons, Brooke D. and Casteels, Kevin R. V. and Edmondson, Edward M. and Fortson, Lucy F. and Kaviraj, Sugata and Keel, William C. and Melvin, Thomas and Nichol, Robert C. and Raddick, M. Jordan and Schawinski, Kevin and Simpson, Robert J. and Skibba, Ramin A. and Smi...

    doi:10.1093/mnras/stt1458 2013

  49. [49]

    Cooke and Jeyhan S

    Kevin C. Cooke and Jeyhan S. Kartaltepe and Caitlin Rose and K. D. Tyler and Behnam Darvish and Sarah K. Leslie and Ying-jie Peng and Boris Häußler and Anton M. Koekemoer , title =. 2022 , month = nov, abstract =

    2022

  50. [50]

    Phillipps and S

    S. Phillipps and S. Bellstedt and M. N. Bremer and R. De Propris and P. A. James and S. Casura and J. Liske and B. W. Holwerda , title =. 2022 , month = nov, abstract =

    2022

  51. [51]

    Skarbinski and Sarah M

    Maya S. Skarbinski and Sarah M. R. Jeffreson and Alyssa A. Goodman , title =. 2022 , month = dec, abstract =

    2022

  52. [52]

    Goldsmith and Laurent Pagani and Tao-Chung Ching and Jinjin Xie and Chen Wang , title =

    Yan Duan and Di Li and Paul F. Goldsmith and Laurent Pagani and Tao-Chung Ching and Jinjin Xie and Chen Wang , title =. 2022 , month = dec, abstract =

    2022

  53. [53]

    Watkins and Ashley Barnes and Kiana F

    Elizabeth J. Watkins and Ashley Barnes and Kiana F. Henny and Hwihyun Kim and Kathryn Kreckel and Sharon E. Meidt and Ralf S. Klessen and Simon C. O. Glover and Thomas G. Williams and B. W. Keller and Adam K. Leroy and Erik W. Rosolowsky and Mederic Boquien and Gagandeep S. Anand and Francesco Belfiore and Frank Bigiel and Guillermo Blanc and Yixian Cao a...

    2022

  54. [54]

    La Vigne and Stuart N

    Misty A. La Vigne and Stuart N. Vogel and Eve C. Ostriker , journal =. 2006 , month = oct, number =. doi:10.1086/506589 , file =

    work page doi:10.1086/506589 2006

  55. [55]

    2022 , month = sep, abstract =

    Francesco Ziparo and Andrea Ferrara and Laura Sommovigo and Mahsa Kohandel , title =. 2022 , month = sep, abstract =

    2022

  56. [56]

    Lee and Karin M

    Janice C. Lee and Karin M. Sandstrom and Adam K. Leroy and David A. Thilker and Eva Schinnerer and Erik Rosolowsky and Kirsten L. Larson and Oleg V. Egorov and Thomas G. Williams and Judy Schmidt and Eric Emsellem and Gagandeep S. Anand and Ashley T. Barnes and Francesco Belfiore and Ivana Beslic and Frank Bigiel and Guillermo A. Blanc and Alberto D. Bola...

    2022

  57. [57]

    Riggi and D

    S. Riggi and D. Magro and R. Sortino and A. De Marco and C. Bordiu and T. Cecconello and A. M. Hopkins and J. Marvil and G. Umana and E. Sciacca and F. Vitello and F. Bufano and A. Ingallinera and G. Fiameni and C. Spampinato and K. Zarb Adami , title =. 2022 , month = dec, abstract =

    2022

  58. [58]

    Trevor Mendel and David R

    Luc Simard and J. Trevor Mendel and David R. Patton and Sara L. Ellison and Alan W. McConnachie , journal =. 2011 , month = aug, number =. doi:10.1088/0067-0049/196/1/11 , groups =

    doi:10.1088/0067-0049/196/1/11 2011

  59. [59]

    Monthly Notices of the Royal Astronomical Society , title =

    Ad. Monthly Notices of the Royal Astronomical Society , title =. 2023 , month = jan, number =. doi:10.1093/mnras/stac3791 , file =

    doi:10.1093/mnras/stac3791 2023

  60. [60]

    and Walmsley, Mike , journal =

    Dickinson, Hugh and Adams, Dominic and Mehta, Vihang and Scarlata, Claudia and Fortson, Lucy and Serjeant, Stephen and Krawczyk, Coleman and Kruk, Sandor and Lintott, Chris and Mantha, Kameswara and Simmons, Brooke D. and Walmsley, Mike , journal =. 2022 , month = oct, number =. doi:10.1093/mnras/stac2919 , file =

    work page doi:10.1093/mnras/stac2919 2022

  61. [61]

    The imprint of clump formation at high redshift. II. The chemistry of the bulge

    Debattista, Victor P. and Liddicott, David J. and Gonzalez, Oscar A. and Silva, Leandro Beraldo e and Amarante, Joao A. S. and Lazar, Ilin and Zoccali, Manuela and Valenti, Elena and Fisher, Deanne B. and Khachaturyants, Tigran and Nidever, David L. and Quinn, Thomas R. and Du, Min and Kassin, Susan , title =. 2023 , month = mar, abstract =. doi:10.48550/...

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2303.08265 2023

  62. [62]

    The imprint of clump formation at high redshift - I

    Adam J Clarke and Victor P Debattista and David L Nidever and Sarah R Loebman and Raymond C Simons and Susan Kassin and Min Du and Melissa Ness and Deanne B Fisher and Thomas R Quinn and James Wadsley and Ken C Freeman and Cristina C Popescu , journal =. The imprint of clump formation at high redshift - I. A disc -abundance dichotomy , year =. doi:10.1093...

    doi:10.1093/mnras/stz104

  63. [63]

    Monthly Notices of the Royal Astronomical Society , title =

    Matteo Messa and Miroslava Dessauges-Zavadsky and Johan Richard and Angela Adamo and David Nagy and Fran. Monthly Notices of the Royal Astronomical Society , title =. 2022 , month = aug, number =. doi:10.1093/mnras/stac2189 , file =

    doi:10.1093/mnras/stac2189 2022

  64. [64]

    Searching for galaxy-scale strong-lenses in galaxy clusters with deep networks -- I: methodology and network performance

    Angora, G. and Rosati, P. and Meneghetti, M. and Brescia, M. and Mercurio, A. and Grillo, C. and Bergamini, P. and Acebron, A. and Caminha, G. and Nonino, M. and Tortorelli, L. and Bazzanini, L. and Vanzella, E. , title =. 2023 , month = mar, abstract =. doi:10.48550/ARXIV.2303.00769 , eprint =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2303.00769 2023

  65. [65]

    Wet Compaction to a Blue Nugget: a Critical Phase in Galaxy Evolution

    Lapiner, Sharon and Dekel, Avishai and Freundlich, Jonathan and Ginzburg, Omri and Jiang, Fangzhou and Kretschmer, Michael and Tacchella, Sandro and Ceverino, Daniel and Primack, Joel , title =. 2023 , month = feb, abstract =. doi:10.48550/ARXIV.2302.12234 , eprint =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2302.12234 2023

  66. [66]

    From Images to Features: Unbiased Morphology Classification via Variational Auto-Encoders and Domain Adaptation

    Xu, Quanfeng and Shen, Shiyin and de Souza, Rafael S. and Chen, Mi and Ye, Renhao and She, Yumei and Chen, Zhu and Ishida, Emille E. O. and Krone-Martins, Alberto and Durgesh, Rupesh , title =. 2023 , month = mar, abstract =. doi:10.48550/ARXIV.2303.08627 , eprint =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2303.08627 2023

  67. [67]

    Genzel and L

    R. Genzel and L. J. Tacconi and F. Eisenhauer and N. M. Förster Schreiber and A. Cimatti and E. Daddi and N. Bouch. Nature , title =. 2006 , month = aug, number =. doi:10.1038/nature05052 , file =

    doi:10.1038/nature05052 2006

  68. [68]

    2006 , month = jul, number =

    The Astrophysical Journal , title =. 2006 , month = jul, number =. doi:10.1086/504403 , file =

    doi:10.1086/504403 2006

  69. [69]

    Clifton Johnson and Anil C

    L. Clifton Johnson and Anil C. Seth and Julianne J. Dalcanton and Matthew L. Wallace and Robert J. Simpson and Chris J. Lintott and Amit Kapadia and Evan D. Skillman and Nelson Caldwell and Morgan Fouesneau and Daniel R. Weisz and Benjamin F. Williams and Lori C. Beerman and Dimitrios A. Gouliermis and Ata Sarajedini , journal =. 2015 , month = apr, numbe...

    doi:10.1088/0004-637x/802/2/127 2015

  70. [70]

    The formation of globular clusters with top-heavy initial mass functions

    Fukushima, Hajime and Yajima, Hidenobu , title =. 2023 , month = mar, abstract =. doi:10.48550/ARXIV.2303.12405 , eprint =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2303.12405 2023

  71. [71]

    An Astronomers Guide to Machine Learning

    Webb, Sara A. and Goode, Simon R. , title =. 2023 , month = apr, abstract =. doi:10.48550/ARXIV.2304.00512 , eprint =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2304.00512 2023

  72. [72]

    Heinzeller and W

    D. Heinzeller and W. J. Duschl , journal =. 2007 , month = jan, number =. doi:10.1111/j.1365-2966.2006.11233.x , groups =

    doi:10.1111/j.1365-2966.2006.11233.x 2007

  73. [73]

    Globular Clusters in the Galactic Center Region: expected behavior in the infalling and merger scenario

    Navarro, Maria Gabriela and Capuzzo-Dolcetta, Roberto and Arca-Sedda, Manuel and Minniti, Dante , journal =. Globular Clusters in the Galactic Center Region: expected behavior in the infalling and merger scenario , year =. doi:10.48550/ARXIV.2303.18123 , eprint =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2303.18123

  74. [74]

    and Tamfal, Tomas and Quinn, Thomas R

    van Donkelaar, Floor and Mayer, Lucio and Capelo, Pedro R. and Tamfal, Tomas and Quinn, Thomas R. and Madau, Piero , title =. 2022 , month = oct, abstract =. doi:10.1093/mnras/stad946 , eprint =

    doi:10.1093/mnras/stad946 2022

  75. [75]

    Elmegreen and Max T

    Debra Meloy Elmegreen and Bruce G. Elmegreen and Max T. Marcus and Karlen Shahinyan and Andrew Yau and Michael Petersen , journal =. 2009 , month = jul, number =. doi:10.1088/0004-637x/701/1/306 , file =

    doi:10.1088/0004-637x/701/1/306 2009

  76. [76]

    Conselice and Jeffrey A

    Christopher J. Conselice and Jeffrey A. Blackburne and Casey Papovich , journal =. The Luminosity, Stellar Mass, and Number Density Evolution of Field Galaxies of Known Morphology from i z /i = 0.5 to 3 , year =. doi:10.1086/426102 , groups =

    doi:10.1086/426102

  77. [77]

    Ferguson and Paolo Cassata and Anton M

    Yicheng Guo and Mauro Giavalisco and Henry C. Ferguson and Paolo Cassata and Anton M. Koekemoer , journal =. 2012 , month = sep, number =. doi:10.1088/0004-637x/757/2/120 , file =

    doi:10.1088/0004-637x/757/2/120 2012

  78. [78]

    Elmegreen and Douglas S

    Debra Meloy Elmegreen and Bruce G. Elmegreen and Douglas S. Rubin and Meredith A. Schaffer , journal =. 2005 , month = sep, number =. doi:10.1086/432502 , groups =

    doi:10.1086/432502 2005

  79. [79]

    Elmegreen and Swara Ravindranath and Daniel A

    Debra Meloy Elmegreen and Bruce G. Elmegreen and Swara Ravindranath and Daniel A. Coe , journal =. 2007 , month = apr, number =. doi:10.1086/511667 , file =

    doi:10.1086/511667 2007

  80. [80]

    YOLO-CL: Galaxy cluster detection in the SDSS with deep machine learning

    Grishin, Kirill and Mei, Simona and Ilic, Stéphane , title =. 2023 , copyright =. doi:10.48550/ARXIV.2301.09657 , groups =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2301.09657 2023

Showing first 80 references.