Selfish Evolution: Making Discoveries in Extreme Label Noise with the Help of Overfitting Dynamics

Ashish Mahabal; Colin Orion Chandler; Maryam Eslami; Nima Sedaghat; Tanawan Chatchadanoraset

REVIEW 3 major objections 4 minor 38 references

Selfish Evolution: Making Discoveries in Extreme Label Noise with the Help of Overfitting Dynamics

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Selfish Evolution uses a model's overfitting history to detect and correct corrupted labels, recovering 817 missed supernovae and cutting MNIST noise from 80% to 21.2%.

desk verdict Overfitting dynamics for label correction is a genuinely fresh idea, but the headline supernova result is likely inflated by a support-batch artifact, and the paper currently lacks the code, data, and controls to back its claims. read the letter →

arxiv 2412.00077 v1 pith:RJ2Q76ZE submitted 2024-11-26 cs.CV astro-ph.IMcs.AIcs.LG

Nima Sedaghat , Tanawan Chatchadanoraset , Colin Orion Chandler , Ashish Mahabal , Maryam Eslami This is my paper · ORCID

classification cs.CVastro-ph.IMcs.AIcs.LG

keywords labelnoiseoverfittingdynamicsevolutioncubeslearningwithnoisylabelssupernovadetectionweaksupervisionimagegenerationclosed-loopcorrection

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 proposes that the sequence of outputs a network produces while being forced to overfit a single sample—its 'overfitting dynamics'—contains enough information to tell whether that sample's label is wrong and what the correct label should be. The authors train a small clean 'gold' subset, record these evolution cubes for each sample, and train a secondary network to map cubes to clean labels, then apply it to the whole noisy training set in a closed loop. On a supernova-detection task with image-like regression labels, the method recovered 817 supernovae that had been missed under 50% label noise; on MNIST with 80% random label corruption, ten closed-loop iterations reduced final noise from 80% to 21.2%. The motivation is that in scientific discovery, a corrupted label is not just noise to be discarded but a potentially missed object worth recovering.

What carries the argument

The key object is the evolution cube: a 3-D tensor stacking the model's outputs on one sample over the overfitting steps, produced by a 'race' between tuning on a clean support batch and overfitting to the sample's current (possibly noisy) label. The Evolution-to-Label (E2L) network, a thinner version of the same encoder-decoder architecture, is trained on gold-subset cubes to map these dynamics directly to clean labels, implicitly performing both noise detection and correction. The closed-loop protocol re-feeds the E2L-corrected labels as the new training labels in successive super-epochs, which allows the system to converge toward a clean dataset.

What would settle it

Run the same protocol on a noisy dataset with known ground truth but with the gold subset drawn from a different distribution than the training set (for instance, a different region of the sky or a different digit style); if cleaned-label accuracy does not beat the baseline without correction, the transfer assumption is false.

Watch

Extended reading notes

Core claim

The central claim is that overfitting dynamics, rather than early-training dynamics, provide a recoverable signal about label noise even when the network is already trained on noisy data. For each sample, starting from a fixed pre-trained state, the network is first tuned on a clean support mini-batch and then driven to overfit the sample itself; the per-epoch outputs form a spatiotemporal evolution cube. The paper shows that these cubes carry enough information that a secondary Evolution-to-Label (E2L) network, trained only on cubes from a small gold subset, can generate corrected labels for the full training set, and that iterating the process makes the labels progressively cleaner without assuming any particular state of the primary network.

Load-bearing premise

The method assumes that the overfitting patterns seen on a small gold subset with clean labels transfer to the much larger noisy training set, so that a correction rule learned from a few hundred examples works across all samples.

Editorial extensions

If this is right

Corrupted labels can be corrected rather than dropped, turning noise into recoverable discoveries in scientific pipelines.
The method works for image-like regression labels (pixel-wise outputs), a regime most label-noise research does not address.
Because it is network-state agnostic, it can be applied to models that have already been fine-tuned for a long time, not just early in training.
Closed-loop iteration produces progressive denoising: ten super-epochs on MNIST cut 80% noise to 21.2%, matching a much larger Co-teaching network.
On the supernova task with 50% noise, the approach recovered 817 missed objects, a direct gain in discovery yield.

Reading between the lines

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

The method may extend to other pixel-level regression tasks, such as medical image segmentation or denoising, where false negatives are costly and a small clean labeled subset exists.
The reliance on a gold subset could potentially be relaxed: if evolution cubes from confident high-loss samples are self-labeled and added to the E2L training set, the method might bootstrap with no clean labels at all.
A testable extension would be to replace the E2L network with a sequence model (RNN or transformer) that reads the temporal dimension explicitly, which might improve correction on longer evolution histories.
The 'race' between support batch and selfish sample suggests the method's success depends on the optimizer's behavior under conflicting gradients; changing optimizer or batch composition could make the signal stronger or weaker, which is a direct empirical knob to probe.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

The core idea is worth taking seriously: instead of early-training dynamics, the authors force the model to overfit individual samples and mine the evolution of outputs for label noise and correction. That is a real shift from Dataset Cartography and similar work, and applying it to image-like labels (astronomical cutouts) rather than classification opens a useful niche. The MNIST experiments, while modest, show the mechanism works at least to some degree. The paper deserves credit for aiming at correction, not just detection, since in the supernova context a missed object is a lost discovery.

The soft spots are substantial. The stress-test concern is valid: the support batch used in the first half of each evolution cube is drawn from the same gold subset G that supplies the E2L training cubes. For a gold sample, that support batch can contain the sample itself, so the clean label leaks into the cube during the first half. E2L can exploit a trivial cue that never exists for the train-set samples. The paper does not exclude the selfish sample from the support batch, and it reports no held-out accuracy on gold cubes. That alone makes the 817-recovered-object headline unreliable. Also, the paper is thin on evidence: three representative runs, no error bars, no code or data, the promised CIFAR supplement absent, and a claim of network-state agnosticism that is never actually tested across different stopping points.

I do not think the method is dead. The MNIST results are less affected by the artifact and show a real signal, so the underlying idea has legs. But as presented, the paper overstates what is proven. A serious referee should ask for code and data, a variant of the supernova experiment where the support batch explicitly excludes the selfish sample, held-out evaluation on gold cubes, and variance across repeated runs. With those, the method could be solid. Without them, the key numbers are not credible.

This is a paper for the LNL and astronomy-ML communities, and it should go to peer review — the novelty and application justify referee time. But I would not publish it in its current form. If the authors can address the contamination and release the artifacts, it could be a useful contribution.

Referee Report

3 major / 4 minor

Summary. The paper proposes Selfish Evolution, a label-noise correction method that works by taking a primary network trained (for any duration) on a noisy dataset, resetting it to a fixed checkpoint, and then generating per-sample 'evolution cubes' by first overfitting to a clean support batch and then to the single sample, recording the network's outputs over epochs. A secondary Evolution-to-Label (E2L) network is trained on evolution cubes from a small clean gold subset, with clean labels as targets, and then applied to the main noisy training set to produce corrected labels; the process can be iterated in closed-loop 'super-epochs'. The method is evaluated on a supernova-detection image-generation task using DESC DC2 cutouts (reporting 817 recovered supernovae at 50% noise) and on MNIST classification with 50% and 80% label noise (reducing final noise from 41.3% to 21.2% after 10 super-epochs, comparable to Co-teaching).

Significance. The core idea is original and potentially valuable: instead of avoiding overfitting, the method explicitly exploits the dynamics of overfitting to individual samples as a signal for detecting and correcting corrupted labels. The application to supernova discovery is scientifically motivated, and the closed-loop scheme for iteratively cleaning a training set is a sensible design. The MNIST comparison to Co-teaching provides a useful sanity check. However, the evidence supporting the central claims is currently incomplete: the headline supernova result comes from a single split with no error bars or specified threshold, the gold/evolution-cube generation has a plausible contamination artifact that is not addressed, and the claimed network-state-agnostic property is never tested. The paper also promises CIFAR results in a supplementary that is not present and provides no code or data, so reproducibility cannot be assessed.

major comments (3)

[§4.1.3–4.1.4] The protocol for generating E2L training cubes does not exclude the selfish sample from the support batch when both are drawn from the gold subset G. Since the support batch is sampled from the clean dataset and the selfish sample for a gold cube is also in G, a cube can contain the same sample's clean image in its first half, leaking the true label through a signal that never occurs for the disjoint training-set samples D used in evaluation. The paper neither excludes this self-overlap nor reports held-out accuracy of E2L on gold cubes (or on D cubes with known clean labels), so the reported 817 recovered objects and 50.1% cosine similarity may be inflated by this train/gold distribution mismatch rather than by the proposed overfitting-dynamics mechanism. Please modify the generation so the support batch excludes the selfish sample, and validate E2L on held-out gold cubes and on D cubes.
[§4.1.5, Table 1] The headline result of 817 recovered supernovae at 50% noise rests on a single run without error bars, and the 'Hard. Discovered objects' column depends on an unspecified threshold for hard similarity. The text states that only 'three main representative' experiments out of several are shown, but the selection protocol is not given, so the reader cannot tell whether 817 is typical, best-case, or a lucky split. Please report mean and standard deviation over multiple seeds, define the hard-similarity and discovery thresholds, and provide results for all experiments rather than a representative subset. This is load-bearing for the paper's central discovery claim.
[§5, Table 3] The claimed 'network-state agnostic' property and 'automatic convergence' of the closed loop are not tested. The method is described as not presuming any particular state of the network (early stopping, fully trained, etc.), but all experiments intervene after an initial training phase of fixed duration, and the MNIST results report only 1 and 10 super-epochs. There is no experiment that interrupts the primary model at different training checkpoints, nor a trajectory of the final noise level over a longer closed-loop run. Without such measurements, the state-agnostic and convergence claims are assertions rather than demonstrated results; please add experiments varying the intervention point and reporting the noise level per super-epoch.

minor comments (4)

[§4.2] The text says CIFAR results are in the supplementary material, but no supplementary material is included with the manuscript; please add it or remove the claim.
[§3, Step 3] The method is described as 'weakly supervised', but E2L is trained on clean gold labels, which is external supervision; please clarify the intended sense of 'weak' or revise the terminology.
[§4.1] There are several presentation issues: 'the idea output' should be 'the ideal output'; Table 2 lacks a caption; the affiliations contain spacing oddities; and 'noised' is used inconsistently with 'noisy'.
[§5] The paper states that source code and data will be made available upon acceptance, but no code or data are provided now; given the centrality of the quantitative claims, providing at least the evaluation protocol and thresholds would improve verifiability.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: E2L is trained on external gold labels and never on its own outputs; the support-batch artifact is a validity concern, not a circular reduction.

full rationale

Walking the derivation chain: E2L is trained on evolution cubes from the gold subset G with clean labels as supervision (Sec. 3 Step 3: phi* = arg min phi ell(g(E_i; phi), y*_i); Sec. 4.1.4: 'E2L is trained on the evolution cubes obtained from the gold dataset'). The corrected labels for D are produced by applying this fixed E2L, and the closed-loop update tilde y_i^(k+1) = g(E_i; phi^(k)) does not feed E2L's own outputs back into E2L's training targets in the reported experiments; each super-epoch retrains E2L on gold. The TransiNet architecture is taken from a self-citation [29], but it is a published base architecture, not the source of the noise-correction claim; no uniqueness theorem or self-imported ansatz is invoked. The paper's own evaluation is on the training set ('we do not evaluate the performance on a clean validation set, but directly on the training set', Sec. 4.1.5), and Sec. 4.1.3 samples the support batch from the clean dataset without excluding the selfish sample, which could let gold cubes contain the clean answer during the support-batch half. These are experimental leakage and evaluation concerns that could inflate the reported numbers, but they are not equation-level reductions of the prediction to its inputs, so they do not constitute circularity. The central derivation is self-contained against external gold supervision and an external MNIST benchmark, so the circularity score is low.

Assumptions & free parameters 5 free parameters · 4 assumptions · 1 invented entities

The method introduces no new physical entities. Its central novelty is a hand-crafted data representation, evolution cubes, plus a set of hyperparameters controlling the overfitting schedule. The core domain assumption, that overfitting dynamics expose the clean label, is not derived but empirically assumed.

free parameters (5)

Number of evolution epochs (Ne) = 60
Set in Section 4.1.3 for the supernova experiment; controls how much overfitting is captured. Not swept.
Support batch learning rate / weight decay / beta1 = 1e-4 / 0.1 / 0.99
Table 2; chosen to keep gradients stable during support-batch tuning.
Selfish learning rate / weight decay / beta1 = 1e-4 / 0 / 0.9
Table 2; chosen to allow overfitting to the selfish sample.
Number of super-epochs (MNIST) = 10
Closed-loop iterations; the paper reports 10-super-epoch runs but does not ablate the number.
Hard similarity threshold = not specified
Used to define discovered objects and hard similarity; threshold value not reported, so the discovery count is not reproducible.

assumptions (4)

domain assumption The sequence of model outputs while overfitting to a single sample carries information about whether the label is correct and what the correct label is.
This is the central premise of Steps 2 and 3; it is not proven or derived, only illustrated in Figures 5 and 6.
domain assumption The gold subset labels are clean and representative enough to train E2L that transfers to the noisy training set.
In Section 4.1.1, 507 gold samples are used; no analysis of gold-to-train distribution shift is given.
domain assumption The method is insensitive to the network state at intervention, despite using a fixed pre-trained state and fixed Ne.
Claimed in Section 1 as network-state agnostic, but Section 4.1.3 always initializes from pre-trained weights; the state is never varied.
ad hoc to paper Standard noise model in Eq (1)-(2) and structured false-negative noise in the supernova task can be treated under the same formulation.
Section 2 presents symmetric instance-independent noise, while Section 4.1 uses missed-object false negatives; the mapping between them is not justified.

invented entities (1)

Evolution cube
purpose: Spatiotemporal tensor of model outputs captured during overfitting, used as input to the E2L correction network.
This is a constructed representation specific to the method, with no externally falsifiable prediction attached.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Selfish Evolution: Making Discoveries in Extreme Label Noise with the Help of Overfitting Dynamics." pith.science (2026). https://pith.science/paper/RJ2Q76ZE

@misc{pith2026241200077,
  author       = {Pith},
  title        = {Pith review of: Selfish Evolution: Making Discoveries in Extreme Label Noise with the Help of Overfitting Dynamics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RJ2Q76ZE}},
  note         = {Machine review of arXiv:2412.00077}
}

read the original abstract

Motivated by the scarcity of proper labels in an astrophysical application, we have developed a novel technique, called Selfish Evolution, which allows for the detection and correction of corrupted labels in a weakly supervised fashion. Unlike methods based on early stopping, we let the model train on the noisy dataset. Only then do we intervene and allow the model to overfit to individual samples. The ``evolution'' of the model during this process reveals patterns with enough information about the noisiness of the label, as well as its correct version. We train a secondary network on these spatiotemporal ``evolution cubes'' to correct potentially corrupted labels. We incorporate the technique in a closed-loop fashion, allowing for automatic convergence towards a mostly clean dataset, without presumptions about the state of the network in which we intervene. We evaluate on the main task of the Supernova-hunting dataset but also demonstrate efficiency on the more standard MNIST dataset.

Figures

Figures reproduced from arXiv: 2412.00077 by the authors.

**Figure 2.** Illustration of various stages of a complete super-epoch. At step 1 (bottom left), the main model is trained on the training subset [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Image-based redefinition of the task of supernova de [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: Results of denoising on one exemplar pair of inputs. The top row is the full image crop, while in the second row, we zoom in to [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

**Figure 5.** Figure 5: Exemplar illustration of a down-sampled, unrolled, evolution cube. The first half (top row) is the first half of the evolution, where [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

**Figure 6.** Figure 6: (left) Evolution histories of some exemplar noised samples in MNIST training set. (middle) Image-like presentation of the [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

**Figure 7.** Figure 7: Exemplar evolution histories of noised data points in the MNIST train set. The first and second columns show that the labels are [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 33 canonical work pages

[1]

The lsst desc dc2 simulated sky survey

Bela Abolfathi, David Alonso, Robert Armstrong, Éric Aubourg, Humna Awan, Yadu N Babuji, Franz Erik Bauer, Rachel Bean, George Beckett, Rahul Biswas, et al. The lsst desc dc2 simulated sky survey. The Astrophysical Journal Supplement Series, 253(1):31,

work page
[2]

Desc dc2 data release note

Bela Abolfathi, Robert Armstrong, Humna Awan, Yadu N Babuji, Franz Erik Bauer, George Beckett, Rahul Biswas, Joanne R Bogart, Dominique Boutigny, Kyle Chard, et al. Desc dc2 data release note. arXiv preprint arXiv:2101.04855, 2021. 5

arXiv 2021
[3]

Image classification with deep learning in the presence of noisy labels: A survey

Gorkem Algan and Ilkay Ulusoy. Image classification with deep learning in the presence of noisy labels: A survey. Knowl. Based Syst., 215:106771, 2019. 2

work page 2019
[4]

A closer look at memorization in deep networks

Devansh Arpit, Stanisław Jastrz˛ ebski, Nicolas Ballas, David Krueger, Emmanuel Bengio, Maxinder S Kan- wal, Tegan Maharaj, Asja Fischer, Aaron Courville, Yoshua Bengio, et al. A closer look at memorization in deep networks. In International conference on machine learning, pages 233–242. PMLR, 2017. 2

work page 2017
[5]

Identifying mis- labeled training data

Carla E Brodley and Mark A Friedl. Identifying mis- labeled training data. Journal of artificial intelligence research, 11:131–167, 1999. 2

work page 1999
[6]

Ai-enhanced citizen science discovery of an ac- tive asteroid:(410590) 2008 gb140

Colin Orion Chandler, Nima Sedaghat, William J Oldroyd, Chadwick A Trujillo, William A Burris, Henry H Hsieh, Jay K Kueny, Kennedy A Farrell, Jarod A DeSpain, Mark Jesus Mendoza Magbanua, et al. Ai-enhanced citizen science discovery of an ac- tive asteroid:(410590) 2008 gb140. Research Notes of the AAS, 8(2):50, 2024. 1

work page 2008
[7]

The active asteroids citizen science program: Overview and first results

Colin Orion Chandler, Chadwick A Trujillo, William J Oldroyd, Jay K Kueny, William A Burris, Henry H Hsieh, Jarod A DeSpain, Nima Sedaghat, Scott S Shep- pard, Kennedy A Farrell, et al. The active asteroids citizen science program: Overview and first results. The Astronomical Journal, 167(4):156, 2024. 1

work page 2024
[8]

Understanding Self-Distillation in the Presence of Label Noise

Rudrajit Das and Sujay Sanghavi. Understanding self-distillation in the presence of label noise. ArXiv, abs/2301.13304, 2023. 2

work page Pith review arXiv 2023

Show all 38 references

[9]

Expertise-based weighting for regression models with noisy labels

Milene Regina dos Santos and Rafael Izbicki. Expertise-based weighting for regression models with noisy labels. ArXiv, abs/2305.07430, 2023. 2

2023 arXiv
[10]

Classification in the presence of label noise: a survey.IEEE transactions on neural networks and learning systems, 25(5):845– 869, 2013

Benoît Frenay and Michel Verleysen. Classification in the presence of label noise: a survey.IEEE transactions on neural networks and learning systems, 25(5):845– 869, 2013. 2

2013
[11]

Classification in the presence of label noise: A survey

Benoît Frénay and Michel Verleysen. Classification in the presence of label noise: A survey. IEEE Transac- tions on Neural Networks and Learning Systems, 25: 845–869, 2014. 2

2014
[12]

Co-teaching: Robust training of deep neu- ral networks with extremely noisy labels

Bo Han, Quanming Yao, Xingrui Yu, Gang Niu, Miao Xu, Weihua Hu, Ivor Wai-Hung Tsang, and Masashi Sugiyama. Co-teaching: Robust training of deep neu- ral networks with extremely noisy labels. In Neural Information Processing Systems, 2018. 3

2018
[13]

Learning from training dynamics: Identifying misla- beled data beyond manually designed features

Qingrui Jia, Xuhong Li, Lei Yu, Jiang Bian, Peng- hao Zhao, Shupeng Li, Haoyi Xiong, and Dejing Dou. Learning from training dynamics: Identifying misla- beled data beyond manually designed features. ArXiv, abs/2212.09321, 2022. 2

2022 arXiv
[14]

Mentornet: Learning data-driven cur- riculum for very deep neural networks on corrupted labels

Lu Jiang, Zhengyuan Zhou, Thomas Leung, Li-Jia Li, and Li Fei-Fei. Mentornet: Learning data-driven cur- riculum for very deep neural networks on corrupted labels. In International Conference on Machine Learn- ing, 2017. 3

2017
[15]

Improving generalization performance by switching from adam to sgd

Nitish Shirish Keskar and Richard Socher. Improving generalization performance by switching from adam to sgd. arXiv preprint arXiv:1712.07628, 2017. 5

2017 arXiv
[16]

Adam: A method for stochastic optimization

Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014. 5

2014 arXiv
[17]

Uncertainty based detection and relabeling of noisy image labels

Jan M Köhler, Maximilian Autenrieth, and William H Beluch. Uncertainty based detection and relabeling of noisy image labels. In CVPR workshops, pages 33–37,
[18]

Learning mul- tiple layers of features from tiny images

Alex Krizhevsky and Geoffrey Hinton. Learning mul- tiple layers of features from tiny images. Technical report, University of Toronto, 2009. 8

2009
[19]

Mnist handwritten digit database, 2010

Yann LeCun and Corinna Cortes. Mnist handwritten digit database, 2010. URL http://yann.lecun. com/exdb/mnist. 8

2010
[20]

Gradient descent with early stopping is provably robust to label noise for overparameterized neural net- works

Mingchen Li, Mahdi Soltanolkotabi, and Samet Oy- mak. Gradient descent with early stopping is provably robust to label noise for overparameterized neural net- works. ArXiv, abs/1903.11680, 2019. 2

1903 arXiv
[21]

Early-learning regu- larization prevents memorization of noisy labels

Sheng Liu, Jonathan Niles-Weed, Narges Razavian, and Carlos Fernandez-Granda. Early-learning regu- larization prevents memorization of noisy labels. Ad- vances in neural information processing systems, 33: 20331–20342, 2020. 2

2020
[22]

Machine learning for the zwicky transient facility

Ashish Mahabal, Umaa Rebbapragada, Richard Wal- ters, Frank J Masci, Nadejda Blagorodnova, Jan van Roestel, Quan-Zhi Ye, Rahul Biswas, Kevin Burdge, Chan-Kao Chang, et al. Machine learning for the zwicky transient facility. Publications of the Astro- nomical Society of the Pac...

2019
[23]

To- wards theoretically understanding why sgd

M Mohammadi, A Mohammadpour, and H Ogata. To- wards theoretically understanding why sgd. In Pro- ceedings of the Conference on Neural Information Pro- cessing Systems, 2020. 5

2020
[24]

Making deep neural networks robust to label noise: A loss correction approach

Giorgio Patrini, Alessandro Rozza, Aditya Krishna Menon, Richard Nock, and Lizhen Qu. Making deep neural networks robust to label noise: A loss correction approach. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 2233–2241,

2017
[25]

Improving data quality with training dynamics of gradient boost- ing decision trees

Moacir Antonelli Ponti, Lucas de Angelis Oliveira, Juan Mart’in Rom’an, and Luis Argerich. Improving data quality with training dynamics of gradient boost- ing decision trees. ArXiv, abs/2210.11327, 2022. 2

2022 arXiv
[26]

Deep learning approach to real-bogus classification for lsst alert production

Nima Sedaghat. Deep learning approach to real-bogus classification for lsst alert production. DMTN 216, LSST Data Management, Jul 2023. URL https: //dmtn-216.lsst.io/. Version 2023-07-10. 2

2023
[27]

Report on the performance of im- age differencing from the perspective of the learning- based classifier task

Nima Sedaghat. Report on the performance of im- age differencing from the perspective of the learning- based classifier task. DMTN 274, LSST, October 2023. URL https://dmtn- 274.lsst.io/DMTN- 274.pdf

2023
[28]

Real-bogus classifier – status report

Nima Sedaghat. Real-bogus classifier – status report. DMTN 272, LSST, April 2024. URL https:// dmtn-272.lsst.io/DMTN-272.pdf. 2

2024
[29]

Effective image differencing with convolutional neural networks for real-time transient hunting

Nima Sedaghat and Ashish Mahabal. Effective image differencing with convolutional neural networks for real-time transient hunting. Monthly Notices of the Royal Astronomical Society, 476(4):5365–5376, 2018. 2, 4, 5

2018
[30]

2016 uu121: An active asteroid discovery via ai- enhanced citizen science

Nima Sedaghat, Colin Orion Chandler, William J Oldroyd, Chadwick A Trujillo, William A Burris, Henry H Hsieh, Jay K Kueny, Kennedy A Farrell, Jarod A DeSpain, Mark Jesus Mendoza Magbanua, et al. 2016 uu121: An active asteroid discovery via ai- enhanced citizen science. Researc...

2016
[31]

Choice over effort: Mapping and diagnosing augmented whole slide image datasets with training dynamics

Wenqi Shi, Benoit Louis Marteau, Felipe Giuste, and May Dongmei Wang. Choice over effort: Mapping and diagnosing augmented whole slide image datasets with training dynamics. In Proceedings of the 14th ACM International Conference on Bioinformatics, Compu- tational Biology, and...
[32]

Learning from noisy labels with deep neural networks: A survey

Hwanjun Song, Minseok Kim, Dongmin Park, Yooju Shin, and Jae-Gil Lee. Learning from noisy labels with deep neural networks: A survey. IEEE transactions on neural networks and learning systems, 2022. 2

2022
[33]

Smith, and Yejin Choi

Swabha Swayamdipta, Roy Schwartz, Nicholas Lourie, Yizhong Wang, Hannaneh Hajishirzi, Noah A. Smith, and Yejin Choi. Dataset cartography: Mapping and diagnosing datasets with training dynamics. In Con- ference on Empirical Methods in Natural Language Processing, 2020. 3

2020
[34]

Joint optimization framework for learning with noisy labels

Daiki Tanaka, Daiki Ikami, Toshihiko Yamasaki, and Kiyoharu Aizawa. Joint optimization framework for learning with noisy labels. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 5552–5560, 2018. 2

2018
[35]

Learning with noisy labels revisited: A study using real-world human annotations

Jiaheng Wei, Zhaowei Zhu, Hao Cheng, Tongliang Liu, Gang Niu, and Yang Liu. Learning with noisy labels revisited: A study using real-world human annotations. ArXiv, abs/2110.12088, 2021. 2

2021 arXiv
[36]

How does disagreement help generalization against label corrup- tion? In International Conference on Machine Learn- ing, 2019

Xingrui Yu, Bo Han, Jiangchao Yao, Gang Niu, Ivor Wai-Hung Tsang, and Masashi Sugiyama. How does disagreement help generalization against label corrup- tion? In International Conference on Machine Learn- ing, 2019. 3

2019
[37]

Generalized cross entropy loss for training deep neural networks with noisy labels

Zhilu Zhang and Mert R Sabuncu. Generalized cross entropy loss for training deep neural networks with noisy labels. In Advances in neural information pro- cessing systems, pages 8778–8788, 2018. 2

2018
[38]

Un- certainty modeling for robust domain adaptation under noisy environments

Junbao Zhuo, Shuhui Wang, and Qingming Huang. Un- certainty modeling for robust domain adaptation under noisy environments. IEEE Transactions on Multimedia,

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

The lsst desc dc2 simulated sky survey

Bela Abolfathi, David Alonso, Robert Armstrong, Éric Aubourg, Humna Awan, Yadu N Babuji, Franz Erik Bauer, Rachel Bean, George Beckett, Rahul Biswas, et al. The lsst desc dc2 simulated sky survey. The Astrophysical Journal Supplement Series, 253(1):31,

work page

[2] [2]

Desc dc2 data release note

Bela Abolfathi, Robert Armstrong, Humna Awan, Yadu N Babuji, Franz Erik Bauer, George Beckett, Rahul Biswas, Joanne R Bogart, Dominique Boutigny, Kyle Chard, et al. Desc dc2 data release note. arXiv preprint arXiv:2101.04855, 2021. 5

arXiv 2021

[3] [3]

Image classification with deep learning in the presence of noisy labels: A survey

Gorkem Algan and Ilkay Ulusoy. Image classification with deep learning in the presence of noisy labels: A survey. Knowl. Based Syst., 215:106771, 2019. 2

work page 2019

[4] [4]

A closer look at memorization in deep networks

Devansh Arpit, Stanisław Jastrz˛ ebski, Nicolas Ballas, David Krueger, Emmanuel Bengio, Maxinder S Kan- wal, Tegan Maharaj, Asja Fischer, Aaron Courville, Yoshua Bengio, et al. A closer look at memorization in deep networks. In International conference on machine learning, pages 233–242. PMLR, 2017. 2

work page 2017

[5] [5]

Identifying mis- labeled training data

Carla E Brodley and Mark A Friedl. Identifying mis- labeled training data. Journal of artificial intelligence research, 11:131–167, 1999. 2

work page 1999

[6] [6]

Ai-enhanced citizen science discovery of an ac- tive asteroid:(410590) 2008 gb140

Colin Orion Chandler, Nima Sedaghat, William J Oldroyd, Chadwick A Trujillo, William A Burris, Henry H Hsieh, Jay K Kueny, Kennedy A Farrell, Jarod A DeSpain, Mark Jesus Mendoza Magbanua, et al. Ai-enhanced citizen science discovery of an ac- tive asteroid:(410590) 2008 gb140. Research Notes of the AAS, 8(2):50, 2024. 1

work page 2008

[7] [7]

The active asteroids citizen science program: Overview and first results

Colin Orion Chandler, Chadwick A Trujillo, William J Oldroyd, Jay K Kueny, William A Burris, Henry H Hsieh, Jarod A DeSpain, Nima Sedaghat, Scott S Shep- pard, Kennedy A Farrell, et al. The active asteroids citizen science program: Overview and first results. The Astronomical Journal, 167(4):156, 2024. 1

work page 2024

[8] [8]

Understanding Self-Distillation in the Presence of Label Noise

Rudrajit Das and Sujay Sanghavi. Understanding self-distillation in the presence of label noise. ArXiv, abs/2301.13304, 2023. 2

work page Pith review arXiv 2023