Pith. sign in

REVIEW 3 major objections 4 minor 43 references

Regularizing Differentiable Architecture Search with Smooth Activation

T0 review · 3 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

Pith's one-line read A smooth-activation regularizer on the architecture weights fixes DARTS's two failure modes—skip-connection collapse and the discretization gap—and yields state-of-the-art classification and super-resolution results.

desk verdict A simple mean-penalty regularizer that may stabilize DARTS in practice, but the theory is asserted, the SOTA claims are overstated, and the dispersing mechanism is never actually explained. read the letter →

arxiv 2504.16306 v1 pith:NRRFYCLZ submitted 2025-04-22 cs.NE cs.CV

classification cs.NEcs.CV
keywords differentiablearchitecturesearchneuralsmoothactivationregularizationskip-connectioncollapsediscretizationgapsuper-resolutionpartialchannel
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

Differentiable architecture search based on DARTS tends to collapse into architectures dominated by skip connections, and the continuous network used during search does not match the discrete network derived at the end. This paper proposes to fix both problems with one auxiliary loss: apply a smooth activation function to the architecture weights $\alpha$ and add that as a regularization term. The loss drives $\alpha$ to large negative values, which makes the softmax operation weights $\beta$ change more gently and removes the unfair advantage of parameter-free operations; its mean-regularizing form then spreads the $\beta$ values apart so the final one-hot choice is made with a clearer margin. The paper reports state-of-the-art accuracy on NAS-Bench-201, CIFAR-10/100, and ImageNet, and better super-resolution models in the DLSR and IMDN search spaces, with search cost cut by the partial-channel variant.

What carries the argument

The load-bearing object is the smooth-activation auxiliary loss $$L_{\mathrm{SA}} = \frac{\lambda_e}{N_o N_e}\sum_{i=1}^{N_e}\sum_{j=1}^{N_o} \frac{(1+\nu)\alpha_j^i + (1-\nu)\alpha_j^i\,\operatorname{erf}(\mu(1-\nu)\alpha_j^i)}{2},$$ whose default case reduces to a mean penalty on $\alpha$. The paper pairs this loss with the identity in Equation (6): for a three-operation search space, $e^{\alpha_{\mathrm{conv}}}+e^{\alpha_{\mathrm{skip}}}+e^{\alpha_{\mathrm{avg}}}$ equals a sum of variances of the candidate outputs around the mixed feature map plus a constant; this identity is what justifies pushing $\alpha$ negative to lower variance and stop skip dominance. The mean-penalty case is then credited with widening the spread of the softmax weights $\beta$, giving a clearer argmax choice and reducing the discretization gap. The same penalty is attached to a partial-channel supernet (SAC-DARTS) to cut memory and search time.

What would settle it

Numerically evaluate both sides of Equation (6) on a real supernet during a NAS-Bench-201 search; if the left and right sides do not agree within tolerance, or the three-argument covariance has no well-defined value, the mechanism is falsified. A behavioral control also works: add a constant to all $\alpha$ under the default mean regularizer; since softmax is shift-invariant, any change in search behavior would show the dispersion effect is not caused by the stated mean-penalty mechanism.

Watch

Extended reading notes

Core claim

The paper's central claim is that the architecture parameters $\alpha$ themselves are the right object to regularize in differentiable architecture search, and that a smooth activation on $\alpha$ solves both skip dominance and discretization discrepancy at once. In a simplified three-operation cell, the paper argues that the skip connection's $\alpha$ is largest because its output stays closest to the optimal feature map, and that driving all $\alpha$ to large negative values reduces the variance of the mixed output and keeps the softmax competition from being decided by exponential amplification. The same loss, in its default mean-regularizing special case, then spreads the $\beta$ values apart in a quasisymmetric way, so the top operation stands out from the rest and the continuous encoding used in search resembles the one-hot encoding used after discretization. Empirically, the paper reports new state-of-the-art results on three classification benchmarks and two super-resolution search spaces, recovery from artificially induced skip dominance, and a flatter validation landscape for the found architecture.

Load-bearing premise

The load-bearing premise is that Equation (6) and its appendix counterparts exactly relate each exponential of the architecture weights to variances and an unspecified three-way covariance of candidate feature maps; the paper asserts these identities with 'it can be easily inferred,' not derived, and if they are false the claimed mechanism for why the new loss stabilizes search is unsupported.

Editorial extensions

If this is right

  • On NAS-Bench-201, SA-DARTS and SAC-DARTS reach the near-optimal 91.55% CIFAR-10 validation accuracy within about 10 search epochs, at a reported cost of 0.8 and 0.4 hours respectively.
  • In the DARTS classification space, SAC-DARTS reports 97.49% top-1 on CIFAR-10, 83.66% on CIFAR-100, and 76.1%/92.9% top-1/top-5 on ImageNet, cutting search time by a factor of four relative to SA-DARTS.
  • In super-resolution, the regularizer improves the DLSR search space on Set5, Set14, B100, and Urban100, and the SAC-searched IMDN variant matches or exceeds IMDN's PSNR with about half the parameters.
  • The regularizer lets the search recover from an initialization that artificially favors skip connections, a condition where both L2 regularization and the Beta (LSE) regularizer remain stuck.
  • The found architectures sit on a flatter validation loss and accuracy landscape, meaning the searched result is less sensitive to perturbations of the architecture parameters.

Reading between the lines

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

  • The loss places no constraint on which operations are preferred, so the same mean-regularizing trick should transfer to any differentiable search that relaxes discrete choices into a softmax, including searches over token choices, pruning masks, or layer widths; this transfer is not tested in the paper.
  • Because softmax is invariant to adding a constant to all $\alpha$, the observed $\beta$ dispersion cannot come from the mean penalty alone; it must arise from an interaction with the optimizer's per-parameter normalization, and making that interaction explicit could turn the method into a more predictable design rule.
  • The loss can be read as an explicit margin-maximizer for architecture selection: it widens the gap between the top softmax weight and the rest. If that reading is right, the same term could replace warm-up schedules and early-stopping rules in other DARTS-style searches.
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

3 major / 4 minor

Summary. The paper proposes SA-DARTS, a regularizer for differentiable architecture search that adds a smooth activation of the architecture weights α as an auxiliary loss; the special case reduces to a mean penalty on α. The authors claim this overcomes skip dominance and discretization discrepancy, and they report state-of-the-art results on NAS-Bench-201, CIFAR-10/100, ImageNet, and super-resolution tasks. A channel-sampled variant, SAC-DARTS, is also introduced. The theoretical justification in Section 2.2 and Appendix 6.4 is meant to show that lowering α reduces variance and that the regularizer spreads the softmax weights β; this is the basis for Claims 1 and 2.

Significance. The proposed regularizer is simple, cheap, and applicable to any differentiable architecture search method, which would be valuable if the mechanism were established. An obvious strength is the breadth of the empirical study: multiple search spaces (DARTS, NAS-Bench-201, DLSR, IMDN), multiple tasks (classification, super-resolution), ablations over hyperparameters, negative initialization experiments, and a comparison with Beta-DARTS and PC-DARTS. However, the theoretical core is asserted rather than derived, and the reported numbers are mostly within noise of previous state of the art, so the abstract's claims of new SOTA results and theoretical support are not currently established. The method may still be useful as a practical regularizer, but the paper's central claims need substantial revision.

major comments (3)
  1. [Section 2.2 and Appendix 6.4, Eqs. (6) and (15)-(17)] These identities are asserted, not derived. The Lagrangian system (12)-(14) yields first-order conditions for β, and the text says 'Following the 2 operator case in [29], we can solve the above equations by recursion,' but no recursion or derivation from (12)-(14) to exponential statements about α is shown. Equation (6) contains an undefined three-argument covariance Cov(x-m(x), oc(x)-m(x), oa(x)-m(x)), and Eqs. (15)-(17) mix m(x) and m* inconsistently. There is also a scaling problem: the left side exp(α) is scale-invariant under rescaling of feature maps, while the variance terms on the right side scale quadratically, so an equality of this form cannot hold without an unspecified, scale-dependent constant. Because Claims 1 and 2 are justified by this analysis, the theoretical mechanism for SA-DARTS is not established.
  2. [Section 3.1, Eq. (8) and Claim 2] The default regularizer in Eq. (8) is a constant shift of every α on every edge in each gradient step under plain gradient descent. Since softmax is shift-invariant, adding the same constant to all logits of an edge leaves β unchanged, so it is not explained how this regularizer increases β dispersion. The paper does not analyze the interaction with Adam's per-parameter normalization, which is the only apparent mechanism that could break the shift invariance. Therefore the claim that the mean penalty 'accommodates larger and smaller β values in a quasisymmetric manner' is not supported by the presented analysis, and Figure 2/Table 1 remain purely empirical demonstrations without a mechanism.
  3. [Abstract, Section 4.2, Section 4.3, Tables 2, 3, 6] The claim of 'new state-of-the-art' results is not supported by the reported numbers. On NAS-Bench-201 (Table 6), SA-DARTS and SAC-DARTS attain exactly the same accuracies as DrNAS and Beta-DARTS on every metric. On CIFAR-10, SAC-DARTS at 97.49±0.04 overlaps with Beta-DARTS at 97.47±0.08; on CIFAR-100, SAC-DARTS at 83.66±0.12 is slightly below Beta-DARTS at 83.76±0.22; on ImageNet, both methods report 76.1 top-1. In Table 3, SA-DLSR is below the IMDN baseline on Set5, Set14, B100, and Urban100, and below the DLSR baseline on Set14. The authors should either temper the SOTA claim or provide a statistical comparison that demonstrates the differences are meaningful.
minor comments (4)
  1. [Section 4.3, sentence after Table 3] The sentence 'As shown in For IMDN search space, ...' is grammatically broken and should be rewritten.
  2. [Table 4 caption] The caption says 'reduced DARTS search space S3' but the table reports results for S1-S4; the caption should refer to all four search spaces.
  3. [Section 6.3] There are typos: 'tune vu' should be 'tune ν' and 'tuning mu' should be 'tuning μ'.
  4. [Section 6.11, Eqs. (21)-(22)] Both equations are labeled E_PC; the second (the ensemble error) should be labeled E_ens for clarity.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the empirical claims are benchmarked externally, and the theoretical identities, while asserted rather than derived, do not reduce to their inputs by construction.

full rationale

The headline results (NAS-Bench-201, CIFAR-10/100, ImageNet, and super-resolution PSNR) are evaluated against external benchmarks and against reproduced baselines in Tables 2, 3, and 6, so the SOTA claims are not fitted targets renamed as predictions. The regularizer in Eqs. 7-8 is defined independently of the benchmarks, and Claims 1-2 are not obtained by fitting a parameter to the reported accuracy values. The supporting theory in Section 2.2 and Appendix 6.4 is indeed asserted rather than derived: Eqs. 3-5 are said to follow by recursion without showing the recursion, Eq. 6 is introduced with 'it can be easily inferred,' and Eqs. 15-17 contain an undefined three-argument covariance and inconsistent use of m(x) versus m*. However, assertion without derivation is a correctness/rigor concern, not circularity, because the paper does not define the regularizer in terms of the claimed effect or use the claimed result as its own input. The citations to the authors' own work (e.g., patent [13] and the future-directions list in Section 7) are not load-bearing for the derivation; no uniqueness theorem or mechanism is imported from a self-citation. The external citation [29] supplies the two-operator case, so it is independent support rather than a self-citation chain. The possible issue that a uniform mean penalty is softmax-invariant and therefore cannot alone explain beta dispersion is a logical gap in the mechanism, not a circular reduction. Accordingly, no circular step can be exhibited, and the appropriate finding is no significant circularity with score 0.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central derivation rests on an asserted variance decomposition (Eq 6 and Appendix Eqs 15-17) that is not proven and contains undefined covariance terms. The claim that the mean penalty disperses β is also assumed. No new entities are introduced; the only additions are loss terms and hyperparameters.

free parameters (3)
  • lambda_e = epoch/5
    Coefficient of the SA loss; grows linearly with epoch and was chosen ad hoc as 'generalizes the best'.
  • nu = 0, 0.25, 1 (per task)
    Slope for negative inputs in the smooth activation; tuned per task as shown in Table 5.
  • mu = 1e6, 1/sqrt(2), 0 (per task)
    Smoothness in the erf term; tuned per task as shown in Table 5 and Appendix 6.3.
assumptions (3)
  • domain assumption Variance decomposition of optimal α (Eq 3-5) from prior work [29] is valid.
    The paper builds its skip dominance argument on this decomposition, which is stated without proof in the simplified 3-operation space.
  • ad hoc to paper Equation 6: sum of exponentials of α equals a linear combination of variances and covariances plus a constant.
    Asserted without derivation; the covariance term with three arguments is undefined and the step from the Lagrangian solution is missing.
  • ad hoc to paper A uniform mean penalty increases β dispersion despite softmax shift invariance.
    The paper does not analyze the mechanism; the claimed effect must come from interaction with Adam normalization, which is never described.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Regularizing Differentiable Architecture Search with Smooth Activation." pith.science (2026). https://pith.science/paper/NRRFYCLZ

@misc{pith2026250416306,
  author       = {Pith},
  title        = {Pith review of: Regularizing Differentiable Architecture Search with Smooth Activation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NRRFYCLZ}},
  note         = {Machine review of arXiv:2504.16306}
}
read the original abstract

Differentiable Architecture Search (DARTS) is an efficient Neural Architecture Search (NAS) method but suffers from robustness, generalization, and discrepancy issues. Many efforts have been made towards the performance collapse issue caused by skip dominance with various regularization techniques towards operation weights, path weights, noise injection, and super-network redesign. It had become questionable at a certain point if there could exist a better and more elegant way to retract the search to its intended goal -- NAS is a selection problem. In this paper, we undertake a simple but effective approach, named Smooth Activation DARTS (SA-DARTS), to overcome skip dominance and discretization discrepancy challenges. By leveraging a smooth activation function on architecture weights as an auxiliary loss, our SA-DARTS mitigates the unfair advantage of weight-free operations, converging to fanned-out architecture weight values, and can recover the search process from skip-dominance initialization. Through theoretical and empirical analysis, we demonstrate that the SA-DARTS can yield new state-of-the-art (SOTA) results on NAS-Bench-201, classification, and super-resolution. Further, we show that SA-DARTS can help improve the performance of SOTA models with fewer parameters, such as Information Multi-distillation Network on the super-resolution task.

Figures

Figures reproduced from arXiv: 2504.16306 by the authors.

Figure 1
Figure 1. Overflow of SAC-DARTS. In addition to the cross-entropy loss, the auxiliary loss [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Comparison between DARTS with L2 regularization ( [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Comparison between different regularization of DARTS: [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: β value changes of the same edge with L2, Beta, and SA on DARTS at an unfair local optimal favoring skip-connection. We assign skip-connection with a higher probability. The first 15 epochs are for warm-up only. Our SAC-DARTS can recover from the unfair disadvantage an…
Figure 4
Figure 4. Figure 4: However, our SAC-DARTS can recover from this extremely unfair competition between the [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: The visualization of validation accuracy and loss landscape with respect to architecture [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: The visualization of DLSR vs. our SAC-DARTS on Urban100 Super-Resolution task (x2). [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Current best normal cell of SAC-DARTS with first 15 epochs on weights only. [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Current best reduction cell of SAC-DARTS with first 15 epochs on weights only. [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: Different regularization methods. If we change [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]
Figure 10
Figure 10. Figure 10: The macro and micro-level supernet during searching phase. The Cell-level Search Space [PITH_FULL_IMAGE:figures/full_fig_p018_10.png]
Figure 11
Figure 11. Figure 11: Improved IMDN model found by SA-DARTS 6.7.3 FLOPs Constraints Therefore, we applied FLOPs constraints on two search techniques to see if they can find smaller models with good performance. We introduce a trade-off search for complexity as an additional loss term in ad…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 31 canonical work pages

  1. [29]

    Re- thinking architecture selection in differentiable nas

    Ruochen Wang, Minhao Cheng, Xiangning Chen, Xiaocheng Tang, and Cho-Jui Hsieh. Re- thinking architecture selection in differentiable nas. arXiv preprint arXiv:2108.04392, 2021

  2. [1]

    Ntire 2017 challenge on single image super-resolution: Dataset and study

    Eirikur Agustsson and Radu Timofte. Ntire 2017 challenge on single image super-resolution: Dataset and study. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops, pages 126–135, 2017

  3. [2]

    Smu: Smooth activation function for deep networks using smoothing maximum technique

    Koushik Biswas, Sandeep Kumar, Shilpak Banerjee, and Ashish Kumar Pandey. Smu: Smooth activation function for deep networks using smoothing maximum technique. arXiv preprint arXiv:2111.04682, 2021

  4. [3]

    Stabilizing differentiable architecture search via perturbation-based regularization

    Xiangning Chen and Cho-Jui Hsieh. Stabilizing differentiable architecture search via perturbation-based regularization. In International conference on machine learning , pages 1554–1565. PMLR, 2020

  5. [4]

    Drnas: Dirichlet neural architecture search

    Xiangning Chen, Ruochen Wang, Minhao Cheng, Xiaocheng Tang, and Cho-Jui Hsieh. Drnas: Dirichlet neural architecture search. arXiv preprint arXiv:2006.10355, 2020

  6. [5]

    Progressive differentiable architecture search: Bridging the depth gap between search and evaluation

    Xin Chen, Lingxi Xie, Jun Wu, and Qi Tian. Progressive differentiable architecture search: Bridging the depth gap between search and evaluation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 1294–1303, 2019

  7. [6]

    Spatiotemporal guided self-supervised depth completion from lidar and monocular camera

    Zhifeng Chen, Hantao Wang, Lijun Wu, Yanlin Zhou, and Dapeng Wu. Spatiotemporal guided self-supervised depth completion from lidar and monocular camera. In 2020 IEEE International conference on visual communications and image processing (VCIP), pages 54–57. IEEE, 2020

  8. [7]

    Darts-: robustly stepping out of performance collapse without indicators

    Xiangxiang Chu, Xiaoxing Wang, Bo Zhang, Shun Lu, Xiaolin Wei, and Junchi Yan. Darts-: robustly stepping out of performance collapse without indicators. arXiv preprint arXiv:2009.01027, 2020

Show all 43 references
  1. [8]

    Fairnas: Rethinking evaluation fairness of weight sharing neural architecture search

    Xiangxiang Chu, Bo Zhang, and Ruijun Xu. Fairnas: Rethinking evaluation fairness of weight sharing neural architecture search. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12239–12248, 2021

  2. [9]

    Fair darts: Eliminating unfair advantages in differentiable architecture search

    Xiangxiang Chu, Tianbao Zhou, Bo Zhang, and Jixiang Li. Fair darts: Eliminating unfair advantages in differentiable architecture search. In European conference on computer vision, pages 465–480. Springer, 2020

  3. [10]

    Distributionally robust produc- tion and replenishment problem for hydrogen supply chains

    Menghua Deng, Bomin Bian, Yanlin Zhou, and Jianpeng Ding. Distributionally robust produc- tion and replenishment problem for hydrogen supply chains. Transportation Research Part E: Logistics and Transportation Review, 179:103293, 2023

  4. [11]

    Stochastic and robust truck-and-drone routing problems with deadlines: A benders decomposition approach

    Menghua Deng, Yuanbo Li, Jianpeng Ding, Yanlin Zhou, and Lianming Zhang. Stochastic and robust truck-and-drone routing problems with deadlines: A benders decomposition approach. Transportation Research Part E: Logistics and Transportation Review, 190:103709, 2024

  5. [12]

    Attentive sensing for efficient multimodal gesture recognition, March 13 2025

    Mostafa El-Khamy, HOR Soheil, and Yanlin Zhou. Attentive sensing for efficient multimodal gesture recognition, March 13 2025. US Patent App. 18/882,626

  6. [13]

    Systems and methods for neural architecture search, February 29 2024

    Mostafa El-Khamy and Yanlin Zhou. Systems and methods for neural architecture search, February 29 2024. US Patent App. 18/148,418

  7. [14]

    Unsupervised learning of depth and camera pose with feature map warping

    Ente Guo, Zhifeng Chen, Yanlin Zhou, and Dapeng Oliver Wu. Unsupervised learning of depth and camera pose with feature map warping. Sensors, 21(3):923, 2021. 11

  8. [15]

    Cm-asap: Cross-modality adaptive sensing and perception for efficient hand gesture recognition

    Soheil Hor, Mostafa El-Khamy, Yanlin Zhou, Amin Arbabian, and SukHwan Lim. Cm-asap: Cross-modality adaptive sensing and perception for efficient hand gesture recognition. In 2024 IEEE 7th International Conference on Multimedia Information Processing and Retrieval (MIPR), pages...

  9. [16]

    Lightweight image super-resolution with hierarchical and differentiable neural architecture search

    Han Huang, Li Shen, Chaoyang He, Weisheng Dong, Haozhi Huang, and Guangming Shi. Lightweight image super-resolution with hierarchical and differentiable neural architecture search. arXiv preprint arXiv:2105.03939, 2021

  10. [17]

    U-darts: Uniform-space differentiable architecture search

    Lan Huang, Shiqi Sun, Jia Zeng, Wencong Wang, Wei Pang, and Kangping Wang. U-darts: Uniform-space differentiable architecture search. Information Sciences, 628:339–349, 2023

  11. [18]

    Lightweight image super-resolution with information multi-distillation network

    Zheng Hui, Xinbo Gao, Yunchu Yang, and Xiumei Wang. Lightweight image super-resolution with information multi-distillation network. In Proceedings of the 27th acm international conference on multimedia, pages 2024–2032, 2019

  12. [19]

    Method and apparatus for learning-based channel matrix prediction, September 10 2024

    Hyukjoon Kwon, Mohamed Mahmoud, Federico Penna, Yanlin Zhou, and Ramy E Ali. Method and apparatus for learning-based channel matrix prediction, September 10 2024. US Patent 12,088,369

  13. [20]

    Visualizing the loss landscape of neural nets

    Hao Li, Zheng Xu, Gavin Taylor, Christoph Studer, and Tom Goldstein. Visualizing the loss landscape of neural nets. Advances in neural information processing systems, 31, 2018

  14. [21]

    Darts+: Improved differentiable architecture search with early stopping

    Hanwen Liang, Shifeng Zhang, Jiacheng Sun, Xingqiu He, Weiran Huang, Kechen Zhuang, and Zhenguo Li. Darts+: Improved differentiable architecture search with early stopping. arXiv preprint arXiv:1909.06035, 2019

  15. [22]

    Darts: Differentiable architecture search

    Hanxiao Liu, Karen Simonyan, and Yiming Yang. Darts: Differentiable architecture search. arXiv preprint arXiv:1806.09055, 2018

  16. [23]

    Retailnet: Enhancing retails of perishable products with multiple selling strategies via pair-wise multi-q learning

    Xiyao Ma, Fan Lu, Xiajun Amy Pan, Yanlin Zhou, and Xiaolin Andy Li. Retailnet: Enhancing retails of perishable products with multiple selling strategies via pair-wise multi-q learning. 2019

  17. [24]

    Improving question generation with sentence- level semantic matching and answer position inferring

    Xiyao Ma, Qile Zhu, Yanlin Zhou, and Xiaolin Li. Improving question generation with sentence- level semantic matching and answer position inferring. In Proceedings of the AAAI conference on artificial intelligence, volume 34, pages 8464–8471, 2020

  18. [25]

    Asking complex questions with multi-hop answer-focused reasoning

    Xiyao Ma, Qile Zhu, Yanlin Zhou, Xiaolin Li, and Dapeng Wu. Asking complex questions with multi-hop answer-focused reasoning. arXiv preprint arXiv:2009.07402, 2020

  19. [26]

    Distributed fault diagnosis with shared-basis and b-splines-based matched learning

    Chen Peng, Yanlin Zhou, and Qing Hui. Distributed fault diagnosis with shared-basis and b-splines-based matched learning. In 2017 13th IEEE Conference on Automation Science and Engineering (CASE), pages 536–541. IEEE, 2017

  20. [27]

    Distributed fault diagnosis of networked dynamical systems with time-varying topology

    Chen Peng, Yanlin Zhou, and Qing Hui. Distributed fault diagnosis of networked dynamical systems with time-varying topology. Journal of the Franklin Institute, 356(11):5754–5780, 2019

  21. [28]

    Server averaging for federated learning

    George Pu, Yanlin Zhou, Dapeng Wu, and Xiaolin Li. Server averaging for federated learning. arXiv preprint arXiv:2103.11619, 2021

  22. [30]

    Snas: stochastic neural architecture search

    Sirui Xie, Hehui Zheng, Chunxiao Liu, and Liang Lin. Snas: stochastic neural architecture search. arXiv preprint arXiv:1812.09926, 2018

  23. [31]

    Partially-connected neural architecture search for reduced computational redun- dancy

    Yuhui Xu, Lingxi Xie, Wenrui Dai, Xiaopeng Zhang, Xin Chen, Guo-Jun Qi, Hongkai Xiong, and Qi Tian. Partially-connected neural architecture search for reduced computational redun- dancy. IEEE Transactions on Pattern Analysis and Machine Intelligence, 43(9):2953–2970, 2021

  24. [32]

    Pc-darts: Partial channel connections for memory-efficient architecture search

    Yuhui Xu, Lingxi Xie, Xiaopeng Zhang, Xin Chen, Guo-Jun Qi, Qi Tian, and Hongkai Xiong. Pc-darts: Partial channel connections for memory-efficient architecture search. arXiv preprint arXiv:1907.05737, 2019

  25. [33]

    b-darts: Beta- decay regularization for differentiable architecture search

    Peng Ye, Baopu Li, Yikang Li, Tao Chen, Jiayuan Fan, and Wanli Ouyang. b-darts: Beta- decay regularization for differentiable architecture search. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10874–10883, 2022. 12

  26. [34]

    Understanding and robustifying differentiable architecture search

    Arber Zela, Thomas Elsken, Tonmoy Saikia, Yassine Marrakchi, Thomas Brox, and Frank Hutter. Understanding and robustifying differentiable architecture search. arXiv preprint arXiv:1909.09656, 2019

  27. [35]

    Theory-inspired path- regularized differential network architecture search

    Pan Zhou, Caiming Xiong, Richard Socher, and Steven Chu Hong Hoi. Theory-inspired path- regularized differential network architecture search. Advances in Neural Information Processing Systems, 33:8296–8307, 2020

  28. [36]

    Design of a distributed real-time e-health cyber ecosystem with collective actions: Diagnosis, dynamic queueing, and decision making

    Yanlin Zhou. Design of a distributed real-time e-health cyber ecosystem with collective actions: Diagnosis, dynamic queueing, and decision making. 2018

  29. [37]

    Developing a dynamic queueing model for the airport check-in process

    Yanlin Zhou, Ryan Anderson, Hamid Vakilzadian, Dietmar PF Moeller, and Andreas Deutschmann. Developing a dynamic queueing model for the airport check-in process. In 2018 IEEE International Conference on Electro/Information Technology (EIT), pages 0871–0876. IEEE, 2018

  30. [38]

    Adaptive leader-follower formation control and obstacle avoidance via deep reinforcement learning

    Yanlin Zhou, Fan Lu, George Pu, Xiyao Ma, Runhan Sun, Hsi-Yuan Chen, and Xiaolin Li. Adaptive leader-follower formation control and obstacle avoidance via deep reinforcement learning. In 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 427...

  31. [39]

    Communication-efficient and attack- resistant federated edge learning with dataset distillation

    Yanlin Zhou, Xiyao Ma, Dapeng Wu, and Xiaolin Li. Communication-efficient and attack- resistant federated edge learning with dataset distillation. IEEE Transactions on Cloud Comput- ing, 11(3):2517–2528, 2022

  32. [40]

    A spiking neural dynamical drift-diffusion model on collective decision making with self-organized criticality

    Yanlin Zhou, Chen Peng, and Qing Hui. A spiking neural dynamical drift-diffusion model on collective decision making with self-organized criticality. In 2018 Annual American Control Conference (ACC), pages 5645–5652. IEEE, 2018

  33. [41]

    Modular platooning and formation control

    Yanlin Zhou, George Pu, Fan Lu, Xiyao Ma, and Xiaolin Li. Modular platooning and formation control

  34. [42]

    Distilled one-shot federated learning

    Yanlin Zhou, George Pu, Xiyao Ma, Xiaolin Li, and Dapeng Wu. Distilled one-shot federated learning. arXiv preprint arXiv:2009.07999, 2020. 13 6 Appendix 6.1 Best Architecture In this section, we present our best architectures searched on CIFAR100. Figure 7: Current best normal...

  35. [43]

    Further, the optimal pair of hyperparameters can be different for different tasks. These two sections of Appendix lead to two conclusions: (i)α values are positively correlated with Var(O(x)−m∗) and negative values ofα help reduce the 3 variance terms on the right side of the ...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.