Pith. sign in

REVIEW 5 major objections 5 minor 16 references

Auto-nnU-Net: Towards Automated Medical Image Segmentation

T0 review · 5 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Automatically searching over training hyperparameters and network architectures improves nnU-Net's segmentation performance on six of ten Medical Segmentation Decathlon datasets and matches it on the rest, at comparable training cost.

desk verdict Useful AutoML wrapper around nnU-Net, but the headline claim only holds against stripped baselines without ensembling or post-processing; worth refereeing after a revised comparison. read the letter →

arxiv 2505.16561 v3 pith:6WUL35GT submitted 2025-05-22 cs.CV cs.AI

Jannis Becktepe , Leona Hennig , Steffen Oeltze-Jafra , Marius Lindauer This is my paper
classification cs.CVcs.AI
keywords automatedmedicalimagesegmentationnnU-NethyperparameteroptimizationneuralarchitecturesearchhierarchicalNASmulti-objectiveDecathlonPriorBand
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

Auto-nnU-Net argues that nnU-Net's fixed hyperparameters and hand-crafted heuristics leave performance on the table, and that treating those choices as searchable variables improves segmentation accuracy. The paper builds an AutoML layer on top of nnU-Net that jointly optimizes training hyperparameters, network architecture, and, in the hierarchical variant, network topology, using Regularized PriorBand to balance accuracy against training runtime. Across all ten Medical Segmentation Decathlon datasets, the resulting configuration beats the convolutional nnU-Net baseline on eight datasets and matches or exceeds all nnU-Net variants on six, with the highest mean test Dice score (77.04%) among the compared methods. A sympathetic reader would care because medical segmentation models are often retrained frequently under tight compute budgets, so a method that finds better models at comparable cost has immediate practical value.

What carries the argument

The load-bearing mechanism is Regularized PriorBand, an extension of the multi-fidelity hyperparameter optimizer PriorBand that turns joint hyperparameter and architecture search into a two-objective problem: minimize $1 - \text{DSC}$ and training runtime. Regularized PriorBand replaces Successive Halving's single-cost selection with non-dominated sorting and crowding-distance ranking, so promising low-budget configurations are promoted toward higher training budgets only when they are accurate, fast, or a useful trade-off; the final incumbent is chosen by accuracy alone. Architecture search is enabled by encoding architectures as hyperparameters, either directly in a unified configuration space or through CFGUNet, which uses a context-free grammar to generate U-Net topologies with variable depth, block counts, normalization, and nonlinearities.

What would settle it

Run the original nnU-Net pipeline, including its default ensemble selection and post-processing, on the same ten MSD datasets and the same 5-fold splits, then compare mean test DSC against the 77.04% reported for Auto-nnU-Net; if full-pipeline nnU-Net's mean matches or exceeds that value, the claim of improving on nnU-Net on 6 of 10 datasets fails on its own comparison.

Watch

Extended reading notes

Core claim

The paper's central claim is that a joint hyperparameter and architecture search can replace nnU-Net's fixed and rule-based settings and improve segmentation performance on most tasks without increasing resource requirements. In 5-fold cross-validation on the Medical Segmentation Decathlon, Auto-nnU-Net outperforms nnU-Net's convolutional 3D U-Net baseline on eight of ten datasets and, on the official test platform, reaches a mean DSC of 77.04%, the highest among nnU-Net baselines, the HPO and HPO+HNAS ablations, and the MedSAM2 foundation model. The paper attributes this to searching over a configuration space that includes optimizer, learning rate, loss, foreground oversampling, augmentation strength, encoder type, feature counts, normalization, and activation, with training runtime as a second objective so that larger or slower models are only adopted when they improve accuracy. It also reports that training hyperparameters such as foreground oversampling and initial learning rate matter more than architectural choices, and that optimized configurations transfer partially across datasets.

Load-bearing premise

The headline comparison pits Auto-nnU-Net against stripped nnU-Net 3D U-Net baselines without post-processing or ensembling, and the paper's own limitations section allows that the original nnU-Net's full pipeline might do better than these stand-ins.

Editorial extensions

If this is right

  • On six of the ten MSD datasets, Auto-nnU-Net's 5-fold cross-validation DSC exceeds every nnU-Net 3D U-Net baseline the paper compared against; on the remaining datasets it is at least on par.
  • The highest mean test DSC of 77.04% on the official MSD leaderboard places Auto-nnU-Net ahead of all reported baselines, including the MedSAM2 foundation model, which trails nnU-Net on nine of ten datasets.
  • Training runtime is an explicit optimization objective, and the Pareto-front analysis shows accuracy-runtime trade-offs on datasets like D03 and D04, with the HPO+HNAS variant cutting training time by a factor of 26 on D04.
  • Hyperparameter-importance analysis (fANOVA) identifies foreground oversampling, initial learning rate, and SGD momentum as consistently influential, suggesting that fixed nnU-Net defaults for these choices are unlikely to be optimal across tasks.
  • Optimized configurations transfer across datasets: the D03 (Liver) incumbent achieves the best DSC on four evaluation datasets, although transfer fails for some pairs such as D04 and D05.

Reading between the lines

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

  • Because training hyperparameters dominate architecture choices in the paper's importance analysis, a cheaper HPO-only search at matched total budget may recover most of the gain; the paper's ablations point in this direction but do not fully isolate it.
  • The cross-dataset transfer matrix suggests that a configuration repository or meta-learned initializer could warm-start future searches, reducing the substantial per-dataset optimization cost the paper reports; that is a natural extension the paper lists as future work.
  • The runtime-aware selection rule matters most in continual or federated settings, where the same model is retrained repeatedly; the paper's Pareto analysis measures single-run savings, so cumulative savings over many retrainings are an untested but plausible consequence.
  • Because MedSAM2 required ground-truth prompts and was excluded from the test leaderboard, the comparison against foundation models is incomplete; a fair test would need a prompt-free variant, which the paper did not have.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper proposes Auto-nnU-Net, a framework that extends nnU-Net with hyperparameter optimization (HPO), neural architecture search (NAS), and hierarchical NAS (HNAS), jointly optimizing segmentation accuracy and training runtime via a new multi-objective method called Regularized PriorBand. The method is evaluated on all ten Medical Segmentation Decathlon (MSD) datasets using 5-fold cross-validation and official test-set submissions, comparing against three stripped nnU-Net 3D U-Net configurations (Conv, ResM, ResL) and a MedSAM2 baseline. The central claim, stated in the abstract and conclusion, is that Auto-nnU-Net 'substantially improves' nnU-Net on 6 of 10 datasets while remaining on par on the others and maintaining practical resource requirements.

Significance. If the central claim were supported, the paper would make a valuable contribution: it introduces a full AutoML pipeline for the widely used nnU-Net framework, extends PriorBand to multi-objective joint architecture and hyperparameter search, and provides an unusually comprehensive evaluation on all ten MSD datasets, including official test-set scores. The authors also release their code, and the study is large-scale (approximately 60k GPU hours). However, the significance is currently limited by a baseline-comparison issue that the authors themselves acknowledge in the Limitations paragraph: all nnU-Net baselines are 3D U-Nets without the ensembling and post-processing components that are integral to the original nnU-Net pipeline. Because the headline claim is framed as improving 'nnU-Net' rather than a stripped variant, the evidence does not yet support it.

major comments (5)
  1. [Section 7, Table 1, Table 6] The central claim that Auto-nnU-Net 'substantially improves the segmentation performance of nnU-Net' is not supported by the baselines used. All nnU-Net baselines are the 3D U-Net without ensembling or post-processing, as explicitly stated in the Limitations paragraph ('This study, including results for both nnU-Net and Auto-nnU-Net, is based on the 3D U-Net architecture without post-processing or ensembling, which may not fully reflect the original nnU-Net's performance'). The original nnU-Net pipeline includes both ensembling and post-processing as default inference components, and these are known to typically increase DSC. The observed improvements over the strongest stripped baseline (Table 1) are small (e.g., D01 +0.24, D02 +0.10, D04 +0.32) and could be erased if the full nnU-Net pipeline recovers even a fraction of a point. To support the abstract's claim, the authors must either add a full nnU-Net baseline (or cite published full-nnU-Net test-set results) or reframe the claim as improving the stripped 3D U-Net component of nnU-Net.
  2. [Abstract, Section 6.1, Table 1, Table 6] The abstract's claim of improvement on '6 out of 10 datasets' is inconsistent with the official test-set results reported in Table 1, where Auto-nnU-Net is the best-performing method on only five datasets (D01, D02, D04, D05, D10); Section 6.1 correctly states 'ranking best on five out of ten datasets.' The '6 out of 10' figure appears to derive from Table 6 (5-fold cross-validation), where Auto-nnU-Net beats the best stripped baseline on D01, D02, D04, D05, D06, and D10. The abstract and main text should report the same numbers, or explicitly distinguish between validation and test-set claims, to avoid overstating the result.
  3. [Section 5.3, Table 1] The optimization is run with 'one random seed' (Section 5.3), and no statistical significance testing is reported. Most improvements over the best stripped baseline in Table 1 are small (e.g., D01 +0.24, D02 +0.10, D04 +0.32) relative to the reported standard deviations (e.g., D01 ±24.3, D10 ±32.0). Without multiple optimization seeds or paired significance tests, the word 'substantially' in the abstract is not statistically justified. The authors should either add a significance analysis (e.g., paired tests across folds or bootstrap intervals) or soften the claim to describe the observed differences without causal or generalizing language.
  4. [Section 5.2, Appendix D.2.1, Table 6] The MedSAM2 baseline is not a fair comparison: it is fine-tuned for only 100 epochs instead of the 1000 used by the original method (Appendix D.2.1), and its inference requires ground-truth-derived bounding-box prompts. The paper acknowledges this, but still labels MedSAM2 as a 'state-of-the-art competitor' in Section 5.2 and uses it in the Pareto-front comparisons. This baseline choice should be clearly disclosed in the main text, and the comparison should be interpreted as indicative rather than as a head-to-head evaluation.
  5. [Section 5 (Introduction of 'practical resource requirements')] The paper emphasizes that Auto-nnU-Net 'maintain[s] practical resource requirements,' but the total search cost is approximately 60,000 GPU hours (Appendix A). The Pareto-front analyses (e.g., Figure 4) refer to training runtime of individual configurations, not the total optimization budget. The authors should specify whether 'practical resource requirements' refers to the inference/training cost of the final selected model or to the total search cost, and discuss the trade-off accordingly.
minor comments (5)
  1. [Section 3 (Related Work)] There is a typo: 'Rule-baesd' should be 'Rule-based'.
  2. [Appendix C.2] The algorithm is referred to as 'Algoritm 2'; please correct to 'Algorithm 2'.
  3. [Table 1 caption] The phrase 'all test set instance DCSs' should be 'DSC scores' or 'Dice scores' for clarity.
  4. [Section 6.1, Figure 2 caption] The caption states 'Detailed results for each dataset are stated in Appendix E,' but the figure is already in the main text; consider referring to specific tables for consistency.
  5. [Appendix D.2.3, Table 4] In the NAS hyperparameter rows, 'Activation' and 'Normalization' are listed as NAS hyperparameters, but they also affect the decoder; clarify the scope of these choices in the text.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the central claim is an empirical comparison on the public MSD test set, not a derivation from fitted inputs or self-cited premises.

full rationale

The paper's central claim is an empirical benchmark result: Auto-nnU-Net's test DSCs on the Medical Segmentation Decathlon are obtained through 5-fold cross-validation and the official MSD test submission platform (Table 1), so the headline improvement over nnU-Net baselines does not reduce by construction to any fitted constant or relabeled prediction. PriorBand and the CFG-based HNAS search space are adopted as off-the-shelf components from cited prior work, and the paper's own contribution (Regularized PriorBand) is presented as a new selection rule in Algorithm 2 rather than as evidence for the benchmark outcome. No load-bearing argument depends on a self-citation chain: the comparison against nnU-Net is against independent training runs. The limitations passage in Section 7 explicitly admits that results are 'based on the 3D U-Net architecture without post-processing or ensembling, which may not fully reflect the original nnU-Net's performance'; this is a real external-validity concern about whether the baselines represent full nnU-Net, but it is a correctness and comparison-adequacy issue, not a circularity issue, because the measured quantities are not themselves defined in terms of the conclusion. No step in the paper's derivation chain equates an output with an input by definition, and no fitted parameter is later presented as a prediction.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No fitted constants are used to derive the central claim; all evaluations use public MSD data and a public optimization framework. The main assumptions are multi-fidelity approximation, validation-to-test transfer, and the adequacy of the simplified nnU-Net baseline, listed above.

assumptions (4)
  • domain assumption Successive Halving evaluations at lower budgets approximate performance at the full 1000-epoch budget.
    Regularized PriorBand relies on multi-fidelity selection in Algorithm 2; the limitation paragraph notes that lower-budget configurations approximate full-budget performance.
  • domain assumption 5-fold cross-validation scores on MSD training splits are predictive of official test-set ranks.
    Configurations are selected on validation scores and then evaluated on the official MSD test platform; no test labels are used during selection.
  • ad hoc to paper The simplified nnU-Net baselines (3D U-Net Conv, ResM, ResL without ensembling or post-processing) are adequate comparators for the claim of improving nnU-Net.
    The Limitations paragraph admits this may not fully reflect the original nnU-Net; since the abstract claims improvement over nnU-Net, the baseline choice is load-bearing.
  • domain assumption Non-dominated sorting with crowding distance selects configurations that balance accuracy and runtime without hurting final accuracy.
    This is central to Regularized PriorBand's selection in Lines 21-25 of Algorithm 2; the final incumbent is picked by the primary objective only.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Auto-nnU-Net: Towards Automated Medical Image Segmentation." pith.science (2026). https://pith.science/paper/6WUL35GT

@misc{pith2026250516561,
  author       = {Pith},
  title        = {Pith review of: Auto-nnU-Net: Towards Automated Medical Image Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6WUL35GT}},
  note         = {Machine review of arXiv:2505.16561}
}
read the original abstract

Medical Image Segmentation (MIS) includes diverse tasks, from bone to organ segmentation, each with its own challenges in finding the best segmentation model. The state-of-the-art AutoML-related MIS-framework nnU-Net automates many aspects of model configuration but remains constrained by fixed hyperparameters and heuristic design choices. As a full-AutoML framework for MIS, we propose Auto-nnU-Net, a novel nnU-Net variant enabling hyperparameter optimization (HPO), neural architecture search (NAS), and hierarchical NAS (HNAS). Additionally, we propose Regularized PriorBand to balance model accuracy with the computational resources required for training, addressing the resource constraints often faced in real-world medical settings that limit the feasibility of extensive training procedures. We evaluate our approach across diverse MIS datasets from the well-established Medical Segmentation Decathlon, analyzing the impact of AutoML techniques on segmentation performance, computational efficiency, and model design choices. The results demonstrate that our AutoML approach substantially improves the segmentation performance of nnU-Net on 6 out of 10 datasets and is on par on the other datasets while maintaining practical resource requirements. Our code is available at https://github.com/automl/AutoNNUnet.

Figures

Figures reproduced from arXiv: 2505.16561 by the authors.

Figure 1
Figure 1. Overview of the Auto-nnU-Net framework: Given a hyperparameter configuration 𝜆, architec￾ture 𝐴, and dataset fingerprint, the AutoExperimentPlanner and CFGUNet generate nnU-Net training plans and model architecture, respectively. The AutoNNUNetTrainer then trains the selected model, providing runtime and validation score as objectives to the PriorBand optimizer. For details, see Appendix C.1. For more details, see A… view at source ↗
Figure 2
Figure 2. Incumbent performance of nnU-Net, Auto-nnU-Net, and Auto-nnU-Net ablations over time. Detailed results for each dataset are stated in Appendix E. Final validation DSCs are stated in [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Qualitative segmentation results for D01. Columns show the input image, ground truth mask, and method predictions, with colors denoting foreground classes. Numbers below method names indicate DSC scores [%] for this example. Each row shows a slice of the 3D volume along one axis. As the 4D volume is an mp-MRI scan, the first parameter setting is used to extract a 3D volume. Additional results are in Appendix E. 20.0… view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Pareto fronts of Auto-nnU-Net and HPO+HNAS for D03 (left) and D04 (right) compared to the baselines and HPO results. Additional results are stated in Appendix E. 6.2 Analysis of Hyperparameter Importance in Auto-nnU-Net To assess the impact of individual hyperparameter…
Figure 5
Figure 5. Figure 5: Overview of the prior-based sampling procedure for HNAS. Each block represents a design decision, i.e., the production rule of the CFG with its corresponding probability distribution. The probability of the possible value is indicated by the blue bar. The probability o…
Figure 6
Figure 6. Figure 6: Example images from the MSD datasets with highlighted target labels, where each color represents an individual class. An image corresponds to the slice with the highest number of foreground voxels in the third image dimension. For 4D volumes, i.e., mp-MRI scans, the fi…
Figure 7
Figure 7. Figure 7: Optimization performance over time. Error bars indicate standard deviation across 5-fold cross-validation splits. 26 [PITH_FULL_IMAGE:figures/full_fig_p026_7.png]
Figure 8
Figure 8. Figure 8: Optimization performance over time. Error bars indicate standard deviation across 5-fold cross-validation splits. 27 [PITH_FULL_IMAGE:figures/full_fig_p027_8.png]
Figure 9
Figure 9. Figure 9: Optimization performance over time. Error bars indicate standard deviation across 5-fold cross-validation splits. Image Ground Truth nnU-Net (Conv) 60.04 nnU-Net (ResM) 68.21 nnU-Net (ResL) 67.99 MedSAM2 34.02 Auto￾nnU-Net 53.12 HPO 60.37 HPO+ HNAS 57.09 Qualitative Re…
Figure 10
Figure 10. Figure 10: Qualitative segmentation results for D01. The columns correspond to the input image, ground truth segmentation mask, and predicted segmentations of the methods, where colors represent foreground classes. Numbers below the method names correspond to their respective DS…
Figure 11
Figure 11. Figure 11: Pareto fronts of HPO+NAS and HPO+HNAS compared to the baselines and HPO results. 29 [PITH_FULL_IMAGE:figures/full_fig_p029_11.png]
Figure 12
Figure 12. Figure 12: Pareto fronts of HPO+NAS and HPO+HNAS compared to the baselines and HPO results. 30 [PITH_FULL_IMAGE:figures/full_fig_p030_12.png]
Figure 13
Figure 13. Figure 13: Cross-evaluation matrix for HPO+NAS incumbent configurations. Each cell states the 5-fold cross-validation DSC [%] when applying an incumbent configuration of a dataset (column) to a different dataset (row). In addition, the mean per incumbent configuration is stated.…
Figure 14
Figure 14. Figure 14: Global functional ANOVA (fANOVA) (Hutter et al., 2014) hyperparameter importance Auto-nnU-Net across all datasets for 1 - DSC with error bars indicating variances. 31 [PITH_FULL_IMAGE:figures/full_fig_p031_14.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

16 extracted references · 12 canonical work pages

  1. [1]

    MO-DEHB: Evolutionary-based Hyperband for Multi-Objective Optimization

    G. J. Aguiar et al. (2019). “A meta-learning approach for selecting image segmentation algorithm”. In: Pattern Recognition Letters. M. J. Ali, M. Essaid, L. Moalic, and L. Idoumghar (2024). “A review of AutoML optimization techniques for medical image applications”. In: Computerized Medical Imaging and Graphics 118, p. 102441. M. Antonelli et al. (2022). ...

    work page Pith review arXiv 2019

  2. [2]

    Since the default for this dataset is four, it is associated with the highest probability

    We begin by sampling the number of stages using its associated production rule, which allows the U-Net to contain two, three, or four stages. Since the default for this dataset is four, it is associated with the highest probability. Here, we consider the network to consist of three stages. Subsequently, the encoder and decoder 20 are sampled. In our examp...

    work page 2019

  3. [10]

    Polynomial Learning Rate Policy with Warm Restart for Deep Neural Network

    P. Mishra and K. Sarawadekar (2019). “Polynomial Learning Rate Policy with Warm Restart for Deep Neural Network”. In: TENCON 2019 - 2019 IEEE Region 10 Conference (TENCON), Kochi, India, October 17-20, 2019 . IEEE. V. Nair and G. Hinton (2010). “Rectified linear units improve restricted boltzmann machines”. In: Proc. of ICML’10. E. Öztürk et al. (2022). “...

    work page 2019

  4. [11]

    For a detailed outline of PriorBand, we refer to Appendix B.1

    are highlighted. For a detailed outline of PriorBand, we refer to Appendix B.1. In Line 21, we apply non-dominated sorting on the set of observations, i.e., candidate configurations𝑃 in the current stage of SH. The subroutine returns a list of fronts, where the first front is the actual Pareto front of𝑃 and each subsequent the updated Pareto front after r...

    work page 2023

  5. [13]

    Adam: A Method for Stochastic Optimization

    D. Kingma and J. Ba (2015). “Adam: A Method for Stochastic Optimization”. In: Proc. of ICLR’15. A. Kirillov et al. (2023). “Segment Anything”. In: Proc. of ICCV’23. L. Li, K. Jamieson, G. DeSalvo, A. Rostamizadeh, and A. Talwalkar (2017). “Hyperband: Bandit-Based Configuration Evaluation for Hyperparameter Optimization”. In: Proc. of ICLR’17. S. Liang et ...

    work page 2015

  6. [14]

    is an image segmentation foundation model pre-trained on a dataset containing 1M images and 1B ground-truth segmentation masks. Unlike task-specific models, e.g., U-Nets, which need to be trained from scratch for each new segmentation dataset, the pre-training enables foundation models to generalize across diverse datasets and reduces the need for extensi...

    work page 2024

  7. [15]

    Segment Anything in Medical Images and Videos: Benchmark and Deployment

    J. Ma et al. (2024b). “Segment Anything in Medical Images and Videos: Benchmark and Deployment”. In: arXiv:2408.03322v1 [eess.IV]. A. L. Maas, A. Y. Hannun, and A. Y. Ng (2013). “Rectifier Nonlinearities Improve Neural Network Acoustic Models”. In: Proc. of ICML’13. N. Mallik et al. (2023). “PriorBand: Practical Hyperparameter Optimization in the Age of D...

    arXiv 2024

  8. [16]

    The video segmentation capabilities of SAM2 enable MedSAM2 to represent 3D volumes as a sequence of 2D frames and produce improved 3D medical image segmentations compared to MedSAM

    and adds a memory attention module to condition the current frame on the previous one. The video segmentation capabilities of SAM2 enable MedSAM2 to represent 3D volumes as a sequence of 2D frames and produce improved 3D medical image segmentations compared to MedSAM. SAM2 was originally designed for interactive image and video segmentation and requires a...

    work page 2023

Show all 16 references
  1. [17]

    (2017) and Mallik et al

    We set the reduction factor𝜂 to the default value of 3 as proposed by Li et al. (2017) and Mallik et al. (2023). We round budgets to full epochs. With the initial evaluation of the default configuration at the maximum fidelity, this leads to 129 evaluated configurations and a ...

    2017

  2. [36]

    Auto-WEKA: combined selection and Hyperparameter Optimization of classification algorithms

    A. L. Simpson et al. (2019).A large annotated medical image dataset for the development and evaluation of segmentation algorithms. D. Stoll et al. (2023). Neural Pipeline Search (NePS) . url: https://github.com/automl/neps. R. Szeliski (2022). Computer Vision - Algorithms and ...

    2019 arXiv

  3. [241]

    Hiera: A Hierarchical Vision Transformer without the Bells-and-Whistles

    C. Ryali et al. (2023). “Hiera: A Hierarchical Vision Transformer without the Bells-and-Whistles”. In: Proc. of ICML’23. N. Schilling, M. Wistuba, and L. Schmidt-Thieme (2016). “Scalable Hyperparameter Optimization with Products of Gaussian Process Experts”. In: Proc. of ECML/...

    2023 arXiv

  4. [340]

    JAHS-Bench-201: A Foundation For Research On Joint Architecture And Hyperparameter Search

    A. Bansal, D. Stoll, M. Janowski, A. Zela, and F. Hutter (2022). “JAHS-Bench-201: A Foundation For Research On Joint Architecture And Hyperparameter Search”. In: Proc. of NeurIPS’22. J. Bergstra and Y. Bengio (2012). “Random Search for Hyper-Parameter Optimization”. In: Journa...

    2022 arXiv

  5. [1135]

    Learning Multiple Defaults for Machine Learning Algorithms

    F. Pfisterer, J. van Rijn, P. Probst, A. Müller, and B. Bischl (2018). “Learning Multiple Defaults for Machine Learning Algorithms”. In: arXiv:1811.09409v3 [stat.ML]. S. Qin et al. (2023). “NG-NAS: Node growth neural architecture search for 3D medical image segmentation”. In: ...

    2018 arXiv

  6. [2015]

    or instance normalization (InstanceNorm) (Ulyanov et al., 2016). 23 Type Hyperparameter Type Range / Values Default Value HPO Optimizer Categorical { SGD, Adam, AdamW} SGD HPO Momentum (SGD) Float (log) [ 0.5, 0.999] 0.99 HPO Initial Learning Rate Float (log) [ 1· 10−5, 0.1] 1...

    2016

  7. [2022]

    is a collection of ten image segmentation datasets from the medical domain. By focusing on diversity with respect to clinical tasks, modalities, and data characteristics, the MSD aims to serve as a standard for the evaluation of image segmentation algorithms. The MSD is public...

  8. [2023]

    addresses this limitation by integrating prior configurations, e.g., expert beliefs, into the optimization process and improves the anytime performance of existing methods such as 𝜋BO (Hvarfner et al., 2022). By incorporating prior knowledge about well- performing regions in t...

    2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.