Pith. sign in

REVIEW 4 major objections 5 minor 37 references

MANAS: Multi-Agent Neural Architecture Search

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read The paper claims that treating each edge of a DARTS cell as an online-learning agent, with shared weights and bandit-style updates, finds better architectures than DARTS at about one-eighth of its memory.

desk verdict A useful multi-agent framing and honest empirical effort, but the central regret guarantee does not cover the actual algorithm and the missing RandomNAS-WS baseline undercuts the empirical claims. read the letter →

arxiv 1909.01051 v4 pith:YEG3IOGS submitted 2019-09-03 cs.CV cs.LGcs.MA

Vasco Lopes , Fabio Maria Carlucci , Pedro M Esperança , Marco Singh , Victor Gabillon , Antoine Yang , Hang Xu , Zewei Chen
show 1 more author
Jun Wang
This is my paper · ORCID
classification cs.CVcs.LGcs.MA MSC 68T07
keywords neuralarchitecturesearchmulti-agentlearningonlineadversarialbanditsweightsharingDARTSspaceCIFAR-10ImageNet
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

The paper claims that neural architecture search can be recast as a multi-agent online-learning problem: one agent per graph edge chooses an operation, all agents share a single set of network weights, and each agent updates its sampling policy from the validation loss alone. Because the agents never materialize all candidate operations at once, the method cuts GPU memory to roughly one eighth of DARTS while searching the final 20-cell network directly instead of searching on 8 cells and stacking them. On CIFAR-10 and ImageNet the method reports lower test error than DARTS and SNAS, and it introduces three new NAS datasets to check that these gains are not just overfitting to CIFAR-10. The paper also proves regret bounds: an exponentially decreasing simple-regret bound for the least-squares variant and an $O(\sqrt{T})$ cumulative-regret bound per agent for the EXP3-style variant. A sympathetic reader would take the central claim to be that per-edge bandit coordination under weight sharing is a viable, memory-cheap way to search deep architectures directly.

What carries the argument

The central object is a per-edge bandit attached to the DARTS cell: agent $A^i$ controls edge $(i,j)$ and chooses one of $K$ operations, and the joint action profile is the architecture $Z_t$. The key identity carrying the argument is the factored regret: under a linear decomposition of the validation loss, the joint worst-case regret splits into a sum of per-agent regrets, letting each agent run an independent bandit in parallel, EXP3 for MANAS and least-squares plus Zipf for MANAS-LS. Weight sharing supplies the common network weights, so no operation parameters are duplicated, which is what gives the memory reduction.

What would settle it

Run MANAS against a uniform-random weight-sharing controller that samples the same architectures and updates the shared weights identically but never updates the sampling distribution, using the same search budget on CIFAR-10 and on the three new datasets; if the random controller matches MANAS's accuracy, the claim that the bandit coordination is what produces the gains is falsified.

Watch

Extended reading notes

Core claim

MANAS is a multi-agent algorithm for neural architecture search in which each directed edge of the DARTS cell is an agent with $K$ possible operations. At every round the agents sample a sparse architecture, the shared weights are updated by backpropagation on the training loss, and the validation loss is used by each agent to update its own estimate of operation quality. Two credit-assignment schemes are proposed: MANAS uses an EXP3-style importance-weighted update with softmax sampling, while MANAS-LS approximates the validation loss as a linear combination of operation contributions and samples with a Zipf distribution. The authors claim that these mechanisms coordinate implicitly through the joint loss, reduce memory from $O(NK)$ to $O(N)$, allow direct search on 20-cell CIFAR-10 and 14-cell ImageNet networks, and deliver test errors of 2.63% and 2.52% on CIFAR-10 and 26.15% on ImageNet, beating DARTS under the same evaluation protocol while using roughly one eighth of its GPU memory.

Load-bearing premise

The regret bound assumes every other agent's actions are fixed for all $T$ rounds, while in the actual algorithm every agent updates every round; if the coupled, non-stationary system is not covered by that assumption, the theoretical guarantee does not apply to MANAS as run.

Editorial extensions

If this is right

  • Search can be run directly on the final network depth, so architecture cells do not need to be repeated; the paper reports gains from this on CIFAR-10.
  • Memory use drops to about one eighth of DARTS, making direct search on ImageNet feasible; the paper reports a 0.85 percentage-point improvement over DARTS and SNAS when searching on ImageNet directly.
  • The regret bounds state that, under the paper's assumptions, cumulative regret grows only as $O(\sqrt{T})$ for MANAS and simple regret decays exponentially for MANAS-LS, so exploration and exploitation are balanced without a hand-tuned reinforcement-learning reward.
  • On three new datasets, Sport-8, Caltech-101 and MIT-67, MANAS outperforms DARTS and random sampling, especially at 14 cells, suggesting the method transfers beyond CIFAR-10.

Reading between the lines

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

  • If the fixed-environment assumption in the regret proof were replaced by a non-stationary or game-theoretic analysis, the same algorithm might enjoy a joint no-regret guarantee; absent that, the theoretical guarantee should be read as per-agent against frozen opponents.
  • The paper's own experiments leave open that a uniform-random weight-sharing controller with the same memory savings could match MANAS's accuracy; a direct head-to-head on the same search budget would isolate the contribution of the bandit updates.
  • Treating the number of cells as a searchable quantity, rather than fixing it, is a natural next step, since the paper finds that the best cell for an 8-cell network is not the best cell for a 14-cell network.
  • Because MANAS needs no differentiable architecture objective, it could in principle be combined with non-differentiable hardware metrics such as latency or energy during search, something gradient-based NAS cannot do directly.
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.

Referee Report

4 major / 5 minor

Summary. The paper frames neural architecture search as a multi-agent online learning problem. Each edge of the DARTS-style cell is controlled by an agent that samples an operation from a private policy, receives the shared validation loss as feedback, and updates its policy. Two instantiations are proposed: MANAS, which uses an EXP3-style softmax update (Eq. 6), and MANAS-LS, which fits a linear model to the validation loss and samples according to a Zipf distribution. The paper claims cumulative regret O(sqrt(T)) for MANAS, exponential simple regret O(e^{-T/H}) for MANAS-LS, a memory reduction to roughly 1/8 of DARTS, state-of-the-art results on CIFAR-10 and ImageNet, and favorable comparisons against random baselines on Sport-8, Caltech-101, and MIT-67. A Gaussian Squeeze Domain experiment is included to illustrate the regret behavior.

Significance. If correct, the manuscript would introduce a genuinely new perspective to NAS: replacing the continuous relaxation of DARTS with a distributed bandit formulation that can search directly over final network depths while retaining theoretical guarantees. The paper deserves credit for several concrete strengths: the agent formulation is clear, the memory-efficient search over 20-cell networks is valuable, the three additional datasets with 8-run means and standard deviations are a step toward reducing benchmark overfitting, and the Gaussian Squeeze Domain experiment makes an explicit empirical connection to the theoretical bound. However, the central advertised guarantees do not hold as stated for the actual algorithm. Theorem 2 is a per-agent bound under a frozen background of other agents, not a bound on the joint regret defined in Eq. (3), and Theorem 1's exponential-rate claim has a scaling problem because H is built from cumulative losses. The empirical evaluation also omits the weight-sharing random-search baseline that would isolate the contribution of the learned updates.

major comments (4)
  1. [§5.3, Appendix D.2, Eq. (3)] Theorem 2 bounds sum_i R*^i_T(pi^{A_i}), where each per-agent regret is defined with a^{-i} fixed for all T rounds. This is not the joint regret in Eq. (3): in Algorithm 1 every agent updates its policy every round, so the loss sequence seen by any one agent is nonstationary because the other agents are also learning. Moreover, summing per-agent regrets with different frozen background action profiles does not compare against min_a sum_t L_t(a), the joint benchmark in Eq. (3). Consequently the abstract's claim of 'vanishing regrets' for the actual coupled dynamics is unsupported. The authors should either prove a bound for the joint dynamics or explicitly restate the contribution as a per-agent guarantee under a stationary background.
  2. [Theorem 1, Appendix D.1] The stated simple-regret bound O(e^{-T/H}) is not justified by the proof. H is defined through the cumulative quantities B_T^{(A_i)}[j] = sum_{t=1}^T beta_t^{(A_i)}[j], so H grows linearly with T for any non-zero mean gap. Then T/H tends to a constant, not to infinity, and O(e^{-T/H}) is not an exponentially decaying guarantee. The proof's exponent, obtained by substituting Delta_k = B_T[k] - B_T[k*_i] into exp(-(Delta_1)^2 T / (2 N log K / lambda_min)), scales as T^3 for stationary gaps, which is not expressed by the claimed T/H form. In addition, the proof assumes a uniform exploration probability 1/log K, while Section 5.1 specifies Zipf sampling; the analysis therefore does not cover the proposed sampler.
  3. [§6.3, Figure 2] The random baseline in Section 6.3 samples eight architectures uniformly and trains each from scratch. This is a random-sampling control, not the weight-sharing random search of Li and Talwalkar (RandomNAS-WS), which is the appropriate baseline for isolating the effect of the bandit updates: RandomNAS-WS has the same memory advantage over DARTS and uses the same weight-sharing protocol, differing only in that it does not update sampling distributions. Without this comparison, the empirical claim that MANAS's learned credit assignment drives the observed gains is not established, and the memory-efficiency claim alone does not differentiate MANAS from the omitted baseline.
  4. [Eq. (5), Eq. (9), Appendix D.1] There is a mismatch between the MANAS-LS algorithm as implemented and the estimator analyzed in the proof. Section 5.1 defines the credit assignment as a least-squares fit over the collected architectures (Eq. 5), while Appendix D.1 analyzes the ComBand-style inverse-propensity estimator ~beta_t = L_t^{(val)} P^dagger Z_t (Eq. 9) under a uniform sampling distribution. These are different estimators, and the paper does not show that the batch least-squares update inherits the regret properties proved for Eq. (9). The theoretical guarantee for MANAS-LS is therefore for a different algorithm than the one evaluated in Section 6.
minor comments (5)
  1. [Algorithm 1, line 1] The initialization says each policy is uniform over j in {1,...,N}, but each agent selects among K operations; this appears to be a typo and should read {1,...,K}.
  2. [Table 1] The reported search costs are internally surprising: 20 cells with 500 epochs cost 2.8 GPU days, while 8 cells with 50 epochs cost 0.8 GPU days. The second has only 1/25 of the search budget, so the costs do not scale as the text would suggest; please clarify whether the numbers are per-run, per-architecture, or otherwise normalized.
  3. [Figure 2] The subplots use different y-axis scales, and on Sport-8 the 14-cell MANAS result is visibly worse than the 8-cell result, while the text states that the 14-cell setting is better. The figure and the discussion should be reconciled, and the authors should offer an explanation for the non-monotonic behavior.
  4. [Section 6.3] The text switches between 'random sampling' and 'random search' without defining the distinction; since the appendix explicitly samples a fixed number of architectures, the terminology should be made consistent.
  5. [Introduction] The sentence 'as show in Section 6 of this work' should read 'as shown in Section 6', and there are several typographical errors elsewhere (e.g., '3 news datasets' and the inconsistent author name 'Xang Xu' vs. 'Hang Xu').

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: results are external-benchmark-based and the theory is a derivation; the only notable self-citation is non-load-bearing, and the main proof gaps are correctness risks, not circularity.

full rationale

The empirical claims (CIFAR-10, ImageNet, Sport-8, Caltech-101, MIT-67) are evaluated against external benchmarks and are not produced by fitting constants or by renaming inputs, so they cannot reduce to the search algorithm's own definitions. The regret analyses are derivations from standard EXP3/ComBand-style arguments (Auer et al. 2002; Cesa-Bianchi and Lugosi 2012) and are not fitted to the observed NAS losses. The Gaussian Squeeze Domain experiment compares the empirical cumulative regret with the derived O(N sqrt(T K log K)) bound rather than defining the result from the bound. The self-citation to Yang et al. (2020), an overlapping-author paper, is used only to corroborate that searching on 8 cells and stacking is a coarse approximation, a point the present paper also demonstrates directly in Section 6; it is not load-bearing. Two genuine weaknesses are visible in the appended material, but they are not circular. First, Theorem 2 in Appendix D.2 is proved only for a per-agent regret with a^{-i} fixed for all T rounds, while Algorithm 1 updates every agent every round, so the claimed joint cumulative regret of Eq. (3) is not established. Second, in Theorem 1, H is defined from cumulative B_T, so O(e^{-T/H}) is not a genuine exponential decay in T as written. These are unsupported-claim and correctness issues, not cases of a prediction being equivalent to its input by construction.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central algorithm rests on the weight-sharing proxy and, for the theory, on either a fixed-other-agents assumption for MANAS or an exact linear loss model for MANAS-LS. The temperature and gamma are tuned on validation data. The paper introduces no new physical or conceptual entities beyond the multi-agent decomposition.

free parameters (2)
  • Softmax temperature eta for MANAS = Tuned on validation; exact value not reported
    Appendix B.1 starts from eta = 0.95 * sqrt(ln(K) / (nK)) and then tunes it to increase validation accuracy during search, so the exploration-exploitation tradeoff is fitted to the validation set.
  • Tuning parameter gamma = Tuned on validation; not defined in the main text
    Appendix B.1 reports gamma = 1.05 * K * ln(K) / n was also tuned, but no gamma appears in the presented sampling policies, leaving the exact algorithm underdetermined.
assumptions (4)
  • domain assumption Shared weights w_t are a valid proxy for w*(Z_t) in the validation loss L_t(Z_t, w_t).
    Algorithm 1 updates weights once per architecture and uses L_val(Z_t, w_{t+1}) as feedback; Appendix E admits this is a simplification and that w_t minimizes E_{Z~pi_t}[L_val(Z, w(Z))] rather than L_val(Z_t, w*(Z_t)). The effect on the regret bounds is not quantified.
  • ad hoc to paper For MANAS-LS, the validation loss satisfies L_s = beta_s^T Z_s for a time-varying beta_s (Eq. 4).
    The least-squares credit assignment and the MANAS-LS regret proof assume exact linearity of the loss in the edge indicator vector; the authors call this crude and provide no justification that it holds for deep-network validation loss.
  • domain assumption beta_t lies in the linear span of sparse architecture vectors, and losses are bounded in [0,1] (Appendix D.1).
    Used to make the estimator tilde beta_t = L_t P^dagger Z_t unbiased and to apply martingale concentration; if the linear model is only approximate, the unbiasedness fails.
  • domain assumption For Theorem 2, the other agents' actions a^{-i} are fixed for T rounds (Appendix D.2).
    This reduces each agent's game to vanilla EXP3, but is violated by the algorithm's simultaneous policy updates.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MANAS: Multi-Agent Neural Architecture Search." pith.science (2026). https://pith.science/paper/YEG3IOGS

@misc{pith2026190901051,
  author       = {Pith},
  title        = {Pith review of: MANAS: Multi-Agent Neural Architecture Search},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YEG3IOGS}},
  note         = {Machine review of arXiv:1909.01051}
}
read the original abstract

The Neural Architecture Search (NAS) problem is typically formulated as a graph search problem where the goal is to learn the optimal operations over edges in order to maximise a graph-level global objective. Due to the large architecture parameter space, efficiency is a key bottleneck preventing NAS from its practical use. In this paper, we address the issue by framing NAS as a multi-agent problem where agents control a subset of the network and coordinate to reach optimal architectures. We provide two distinct lightweight implementations, with reduced memory requirements (1/8th of state-of-the-art), and performances above those of much more computationally expensive methods. Theoretically, we demonstrate vanishing regrets of the form O(sqrt(T)), with T being the total number of rounds. Finally, aware that random search is an, often ignored, effective baseline we perform additional experiments on 3 alternative datasets and 2 network configurations, and achieve favourable results in comparison.

Figures

Figures reproduced from arXiv: 1909.01051 by the authors.

Figure 1
Figure 1. MANAS with single cell. Between each pair of nodes, an agent Ai selects action a (i) according to π (i) . Feedback from the validation loss is used to update the policy. NAS suffers from a combinatorial explosion in its search space. A recently proposed ap￾proach to tackle this problem is to approxi￾mate the discrete optimisation variables (i.e., edges in our case) with continuous counterparts and then use gradient-… view at source ↗
Figure 2
Figure 2. Comparing MANAS, random sampling and DARTS (Liu et al., 2019) on [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Regret for the Gaussian Squeeze Domain experiment with [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Theoretical bound for the MANAS cumulative regret ( [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 26 canonical work pages

  1. [1]

    Best of both worlds: Stochastic & adversarial best-arm identification

    Yasin Abbasi-Yadkori, Peter Bartlett, Victor Gabillon, Alan Malek, and Michal Valko. Best of both worlds: Stochastic & adversarial best-arm identification. In Conference on Learning Theory (COLT), 2018

    work page 2018

  2. [2]

    The nonstochastic multiarmed bandit problem

    Peter Auer, Nicolo Cesa--Bianchi , Yoav Freund, and Robert E Schapire. The nonstochastic multiarmed bandit problem. SIAM journal on computing, 32 0 (1): 0 48--77, 2002

    work page 2002

  3. [3]

    Regret analysis of stochastic and nonstochastic multi-armed bandit problems

    S \'e bastien Bubeck, Nicolo Cesa-Bianchi, et al. Regret analysis of stochastic and nonstochastic multi-armed bandit problems. Foundations and Trends in Machine Learning , 5 0 (1): 0 1--122, 2012

    2012

  4. [4]

    Proxyless NAS : Direct neural architecture search on target task and hardware

    Han Cai, Ligeng Zhu, and Song Han. Proxyless NAS : Direct neural architecture search on target task and hardware. In International Conference on Learning Representations (ICLR), 2019

    work page 2019

  5. [5]

    Combinatorial bandits

    Nicolo Cesa-Bianchi and G \'a bor Lugosi. Combinatorial bandits. Journal of Computer and System Sciences, 78 0 (5): 0 1404--1422, 2012

    2012

  6. [6]

    Counterfactual exploration for improving multiagent learning

    Mitchell K Colby, Sepideh Kharaghani, Chris HolmesParker, and Kagan Tumer. Counterfactual exploration for improving multiagent learning. In Autonomous Agents and Multiagent Systems (AAMAS 2015), pages 171--179. International Foundation for Autonomous Agents and Multiagent Systems, 2015

    work page 2015

  7. [7]

    Very deep convolutional networks for text classification

    Alexis Conneau, Holger Schwenk, Lo \" c Barrault, and Yann Lecun. Very deep convolutional networks for text classification. In European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pages 1107--1116, 2017

    work page 2017

  8. [8]

    Autoaugment: Learning augmentation policies from data

    Ekin D Cubuk, Barret Zoph, Dandelion Mane, Vijay Vasudevan, and Quoc V Le. Autoaugment: Learning augmentation policies from data. arXiv:1805.09501, 2018

    arXiv 2018

Show all 37 references
  1. [9]

    ImageNet : A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. ImageNet : A large-scale hierarchical image database. In Computer Vision and Pattern Recognition (CVPR), pages 248--255, 2009

    2009

  2. [10]

    Learning generative visual models from few training examples: An incremental bayesian approach tested on 101 object categories

    Li Fei-Fei, Rob Fergus, and Pietro Perona. Learning generative visual models from few training examples: An incremental bayesian approach tested on 101 object categories. Computer Vision and Image Understanding, 106 0 (1): 0 59--70, 2007

    2007

  3. [11]

    Freedman

    David A. Freedman. On tail probabilities for martingales. The Annals of Probability, pages 100--118, 1975

    1975

  4. [12]

    Deep pyramidal residual networks

    Dongyoon Han, Jiwhan Kim, and Junmo Kim. Deep pyramidal residual networks. In Computer Vision and Pattern Recognition (CVPR), pages 5927--5935, 2017

    2017

  5. [13]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Computer Vision and Pattern Recognition (CVPR), pages 770--778, 2016

    2016

  6. [14]

    Densely connected convolutional networks

    Gao Huang, Zhuang Liu, Laurens Van Der Maaten, and Kilian Q Weinberger. Densely connected convolutional networks. In Computer Vision and Pattern Recognition (CVPR), pages 4700--4708, 2017

    2017

  7. [15]

    Imagenet classification leaderboard

    ByungSoo Ko. Imagenet classification leaderboard. https://kobiso.github.io/Computer-Vision-Leaderboard/imagenet, 2019

    2019

  8. [16]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky. Learning multiple layers of features from tiny images. Technical report, University of Toronto, 2009

    2009

  9. [17]

    What, where and who? classifying events by scene and object recognition

    Li-Jia Li and Li Fei-Fei. What, where and who? classifying events by scene and object recognition. In International Conference on Computer Vision (ICCV), pages 1--8, 2007

    2007

  10. [18]

    Random search and reproducibility for neural architecture search

    Liam Li and Ameet Talwalkar. Random search and reproducibility for neural architecture search. arXiv:1902.07638, 2019

    1902 arXiv

  11. [19]

    Progressive neural architecture search

    Chenxi Liu, Barret Zoph, Maxim Neumann, Jonathon Shlens, Wei Hua, Li-Jia Li, Li Fei-Fei, Alan Yuille, Jonathan Huang, and Kevin Murphy. Progressive neural architecture search. In European Conference on Computer Vision (ECCV), pages 19--34, 2018 a

    2018

  12. [20]

    Hierarchical representations for efficient architecture search

    Hanxiao Liu, Karen Simonyan, Oriol Vinyals, Chrisantha Fernando, and Koray Kavukcuoglu. Hierarchical representations for efficient architecture search. In International Conference on Learning Representations (ICLR), 2018 b

    2018

  13. [21]

    DARTS : Differentiable architecture search

    Hanxiao Liu, Karen Simonyan, and Yiming Yang. DARTS : Differentiable architecture search. In International Conference on Learning Representations (ICLR), 2019

    2019

  14. [22]

    Regularizing and optimizing LSTM language models

    Stephen Merity, Nitish Shirish Keskar, and Richard Socher. Regularizing and optimizing LSTM language models. In International Conference on Learning Representations (ICLR), 2018

    2018

  15. [23]

    Efficient neural architecture search via parameter sharing

    Hieu Pham, Melody Guan, Barret Zoph, Quoc Le, and Jeff Dean. Efficient neural architecture search via parameter sharing. In International Conference on Machine Learning (ICML), pages 4092--4101, 2018

    2018

  16. [24]

    Recognizing indoor scenes

    Ariadna Quattoni and Antonio Torralba. Recognizing indoor scenes. In Computer Vision and Pattern Recognition (CVPR), pages 413--420, 2009

    2009

  17. [25]

    QMIX : Monotonic value function factorisation for deep multi-agent reinforcement learning

    Tabish Rashid, Mikayel Samvelyan, Christian Schroeder Witt, Gregory Farquhar, Jakob Foerster, and Shimon Whiteson. QMIX : Monotonic value function factorisation for deep multi-agent reinforcement learning. In International Conference on Machine Learning (ICML), pages 4292--4301, 2018

    2018

  18. [26]

    Large-scale evolution of image classifiers

    Esteban Real, Sherry Moore, Andrew Selle, Saurabh Saxena, Yutaka Leon Suematsu, Jie Tan, Quoc V Le, and Alexey Kurakin. Large-scale evolution of image classifiers. In International Conference on Machine Learning (ICML), pages 2902--2911, 2017

    2017

  19. [27]

    Regularized evolution for image classifier architecture search

    Esteban Real, Alok Aggarwal, Yanping Huang, and Quoc V Le. Regularized evolution for image classifier architecture search. arXiv:1802.01548, 2018

    2018 arXiv

  20. [28]

    Evaluating the search phase of neural architecture search

    Christian Sciuto, Kaicheng Yu, Martin Jaggi, Claudiu Musat, and Mathieu Salzmann. Evaluating the search phase of neural architecture search. arXiv:1902.08142, 2019

    1902 arXiv

  21. [29]

    Inception-v4, Inception-ResNet and the impact of residual connections on learning

    Christian Szegedy, Sergey Ioffe, Vincent Vanhoucke, and Alexander A Alemi. Inception-v4, Inception-ResNet and the impact of residual connections on learning. In AAAI Conference on Artificial Intelligence, 2017

    2017

  22. [30]

    SNAS : Stochastic neural architecture search

    Sirui Xie, Hehui Zheng, Chunxiao Liu, and Liang Lin. SNAS : Stochastic neural architecture search. In International Conference on Learning Representations (ICLR), 2019

    2019

  23. [31]

    NAS evaluation is frustratingly hard

    Antoine Yang, Pedro M Esperan c a, and Fabio M Carlucci. NAS evaluation is frustratingly hard. In International Conference on Learning Representations (ICLR), 2020

    2020

  24. [32]

    ShuffleNet : An extremely efficient convolutional neural network for mobile devices

    Xiangyu Zhang, Xinyu Zhou, Mengxiao Lin, and Jian Sun. ShuffleNet : An extremely efficient convolutional neural network for mobile devices. In Conference on Computer Vision and Pattern Recognition (CVPR), pages 6848--6856, 2018

    2018

  25. [33]

    Neural architecture search with reinforcement learning

    Barret Zoph and Quoc Le. Neural architecture search with reinforcement learning. In International Conference on Learning Representations (ICLR), 2017

    2017

  26. [34]

    Learning transferable architectures for scalable image recognition

    Barret Zoph, Vijay Vasudevan, Jonathon Shlens, and Quoc V Le. Learning transferable architectures for scalable image recognition. In Computer Vision and Pattern Recognition (CVPR), pages 8697--8710, 2018

    2018

  27. [35]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should ...

  28. [36]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@firs...

  29. [37]

    linearity of convolution

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibset...

    2008

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.