Survival of the Cheapest: Cost-Aware Hardware Adaptation for Adversarial Robustness
Pith reviewed 2026-05-23 21:12 UTC · model grok-4.3
The pith
Cheaper GPUs can increase a model's survival time against attacks by 20 percent while cutting costs by 75 percent.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The paper establishes that accelerated failure time models applied to adversarial survival time can quantify the impact of hardware selection, batch size, epochs, and validation accuracy, with results showing the Nvidia L4 delivering 20 percent longer survival time at 75 percent lower cost than the V100 and inference latency emerging as the dominant predictor over training duration or hardware configuration.
What carries the argument
Accelerated failure time models that treat adversarial survival time as the outcome variable driven by hardware, batch size, epochs, and validation accuracy.
If this is right
- Hardware and hyperparameter choices can be ranked directly by predicted survival time to favor cheaper configurations that still raise robustness.
- Inference latency becomes the primary metric for deciding where to run models, shifting attention from training hardware to deployment speed.
- Continuous monitoring of cost, robustness, and latency can drive automatic selection of new hardware or settings as conditions change.
Where Pith is reading between the lines
- The same survival modeling could be tested on non-GPU accelerators or CPU-only setups to see whether the cost-robustness pattern holds.
- Optimizing explicitly for inference latency might yield robustness gains even without changing hardware, a direct extension worth checking.
- The approach could apply to measuring resilience against other threats such as data poisoning if survival time is redefined accordingly.
Load-bearing premise
Survival time until a model fails under adversarial attack is a valid and sufficient measure of robustness, and the models accurately capture how hardware and other factors change that time.
What would settle it
Run the same adversarial attacks on models trained and deployed on L4 versus V100 GPUs and check whether the measured failure times match the survival times predicted by the models.
read the original abstract
Deploying adversarially robust machine learning systems requires continuous trade-offs between robustness, cost, and latency. We present an autonomic decision-support framework providing a quantitative foundation for adaptive hardware selection and hyper-parameter tuning in cloud-native deep learning. The framework applies accelerated failure time (AFT) models to quantify the effect of hardware choice, batch size, epochs, and validation accuracy on model survival time. This framework can be naturally integrated into an autonomic control loop (monitor--analyse--plan--execute, MAPE-K), where system metrics such as cost, robustness, and latency are continuously evaluated and used to adapt model configurations and hardware selection. Experiments across three GPU architectures confirm the framework is both sound and cost-effective: the Nvidia L4 yields a 20% increase in adversarial survival time while costing 75% less than the V100, demonstrating that expensive hardware does not necessarily improve robustness. The analysis further reveals that model inference latency is a stronger predictor of adversarial robustness than training time or hardware configuration.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proposes an autonomic decision-support framework that applies accelerated failure time (AFT) models to quantify how hardware choice, batch size, epochs, and validation accuracy affect 'adversarial survival time' as a robustness measure. It claims experiments across three GPU architectures demonstrate the framework is sound and cost-effective, with the Nvidia L4 yielding a 20% increase in adversarial survival time at 75% lower cost than the V100; inference latency is reported as a stronger predictor of robustness than training time or hardware configuration. The framework is positioned for integration into MAPE-K control loops for adaptive hardware and hyper-parameter selection.
Significance. If the adversarial survival time metric were shown to correlate with established robustness measures, the work could provide a practical, quantitative basis for cost-aware hardware adaptation in robust ML deployments, with the latency-predictor finding offering actionable system-design guidance. The cost-effectiveness demonstration (L4 vs. V100) would then carry weight for cloud-native settings. At present the significance is constrained by the unvalidated metric.
major comments (3)
- [Abstract] Abstract: The central claim that the L4 hardware yields a 20% increase in adversarial survival time (hence better robustness) at 75% lower cost than the V100 rests on survival time being a meaningful proxy for adversarial robustness. The manuscript supplies no evidence that this metric correlates with or substitutes for standard measures such as robust accuracy under PGD/FGSM attacks or attack success rates; without this link the hardware-adaptation conclusions cannot be interpreted as robustness improvements.
- [Abstract] Abstract: The AFT models are asserted to quantify the effects of hardware, batch size, epochs, and validation accuracy on survival time and to identify inference latency as the strongest predictor, yet no details are provided on the AFT functional form, the definition of the failure event, censoring treatment, coefficient estimation, or any goodness-of-fit or significance tests. These omissions make it impossible to verify the reported 20% improvement or the predictor ranking.
- [Abstract] Abstract: The experimental claim of a 20% survival-time advantage for the L4 is presented without error bars, replication counts, statistical tests, or model specifications (architecture, attack parameters, training details). This prevents assessment of whether the difference is reliable or whether the three-GPU comparison supports the broader conclusion that expensive hardware does not necessarily improve robustness.
Simulated Author's Rebuttal
We thank the referee for the constructive feedback on our manuscript. The comments highlight important aspects of metric validation and methodological transparency that we will address in revision. Below we respond point-by-point to the major comments.
read point-by-point responses
-
Referee: [Abstract] The central claim that the L4 hardware yields a 20% increase in adversarial survival time at 75% lower cost than the V100 rests on survival time being a meaningful proxy for adversarial robustness. The manuscript supplies no evidence that this metric correlates with or substitutes for standard measures such as robust accuracy under PGD/FGSM attacks or attack success rates.
Authors: We agree that demonstrating a link between adversarial survival time and established robustness metrics would strengthen interpretability. Adversarial survival time measures time-to-failure under sustained attack and is therefore conceptually related to robustness, yet we did not include an explicit correlation study. In the revised manuscript we will add a dedicated subsection reporting Pearson and Spearman correlations between survival time and robust accuracy (PGD, epsilon=0.03) across the evaluated configurations, together with scatter plots. revision: yes
-
Referee: [Abstract] The AFT models are asserted to quantify the effects of hardware, batch size, epochs, and validation accuracy on survival time and to identify inference latency as the strongest predictor, yet no details are provided on the AFT functional form, the definition of the failure event, censoring treatment, coefficient estimation, or any goodness-of-fit or significance tests.
Authors: The full manuscript (Section 3) specifies a log-normal AFT model, defines the failure event as the first epoch at which adversarial accuracy falls below 50 percent under the chosen attack, treats right-censored observations for runs that reached the maximum experiment duration, and reports coefficient estimates with p-values. To improve accessibility we will expand this section with the explicit survival function, the estimation procedure (maximum likelihood via survreg), and model diagnostics including the concordance index and residual plots. revision: partial
-
Referee: [Abstract] The experimental claim of a 20% survival-time advantage for the L4 is presented without error bars, replication counts, statistical tests, or model specifications (architecture, attack parameters, training details).
Authors: The 20 percent figure is the AFT-predicted multiplicative effect for the L4 versus V100 after controlling for batch size and epochs. We will augment the results section with bootstrap-derived 95 percent confidence intervals on the survival-time ratios, the number of independent training runs per hardware-batch combination (five), and the results of likelihood-ratio tests comparing nested AFT models. Attack parameters (PGD, 20 iterations, epsilon=0.03) and model architecture (ResNet-18) are already stated in the experimental setup; we will ensure they are also summarized in the abstract revision. revision: yes
Circularity Check
No significant circularity; claims rest on empirical AFT fitting to measured data
full rationale
The paper fits AFT models to experimentally measured adversarial survival times across GPU architectures, batch sizes, epochs, and validation accuracies, then reports direct comparisons (e.g., L4 vs V100 survival time and cost). No step reduces a claimed prediction or result to its own fitted inputs by construction, nor relies on self-citation chains, uniqueness theorems from the authors, or smuggled ansatzes. Survival time is treated as an observed quantity whose effects are quantified by regression; the derivation chain remains independent of the target conclusions.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption Accelerated failure time models are suitable for modeling adversarial model robustness as a survival-time outcome dependent on hardware and training variables.
invented entities (1)
-
adversarial survival time
no independent evidence
Lean theorems connected to this paper
-
IndisputableMonolith/Cost/FunctionalEquation.leanwashburn_uniqueness_aczel unclear?
unclearRelation between the paper passage and the cited Recognition theorem.
We propose using accelerated failure time models to measure the effect of hardware choice, batch size, number of epochs, and test-set accuracy by using adversarial attacks to induce failures
-
IndisputableMonolith/Foundation/RealityFromDistinction.leanreality_from_one_distinction unclear?
unclearRelation between the paper passage and the cited Recognition theorem.
the Nvidia L4 yields a 20% increase in adversarial survival time while costing 75% less than the V100
What do these tags mean?
- matches
- The paper's claim is directly supported by a theorem in the formal canon.
- supports
- The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
- extends
- The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
- uses
- The paper appears to rely on the theorem as machinery.
- contradicts
- The paper's claim conflicts with a theorem or certificate in the canon.
- unclear
- Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.
Reference graph
Works this paper leans on
-
[1]
Radiographics 37(2), 505–515 (2017) 23
Erickson, B.J., Korfiatis, P., Akkus, Z., Kline, T.L.: Machine learning for medical imaging. Radiographics 37(2), 505–515 (2017) 23
work page 2017
-
[2]
In: 2018 Aviation Technology, Integration, and Operations Conference, p
Maheshwari, A., Davendralingam, N., DeLaurentis, D.A.: A comparative study of machine learning techniques for aviation applications. In: 2018 Aviation Technology, Integration, and Operations Conference, p. 3980 (2018)
work page 2018
-
[3]
In: 31st USENIX Security Symposium (USENIX Security 22), pp
Arp, D., Quiring, E., Pendlebury, F., Warnecke, A., Pierazzi, F., Wressnegger, C., Cavallaro, L., Rieck, K.: Dos and don’ts of machine learning in computer security. In: 31st USENIX Security Symposium (USENIX Security 22), pp. 3971–3988 (2022)
work page 2022
-
[4]
IEEE Transactions on Systems, Man, and Cybernetics: Systems 47(4), 682–692 (2016)
Mery, D., Svec, E., Arias, M., Riffo, V., Saavedra, J.M., Banerjee, S.: Modern com- puter vision techniques for x-ray testing in baggage inspection. IEEE Transactions on Systems, Man, and Cybernetics: Systems 47(4), 682–692 (2016)
work page 2016
-
[5]
International journal of environmental research and public health 19(17), 10594 (2022)
Travaini, G.V., Pacchioni, F., Bellumore, S., Bosia, M., De Micco, F.: Machine learning and criminal justice: A systematic review of advanced methodology for recidivism risk prediction. International journal of environmental research and public health 19(17), 10594 (2022)
work page 2022
-
[6]
End to End Learning for Self-Driving Cars
Bojarski, M., Del Testa, D., Dworakowski, D., Firner, B., Flepp, B., Goyal, P., Jackel, L.D., Monfort, M., Muller, U., Zhang, J., et al.: End to end learning for self-driving cars. arXiv preprint arXiv:1604.07316 (2016)
work page internal anchor Pith review Pith/arXiv arXiv 2016
-
[7]
IEEE transactions on neural networks 10(5), 988–999 (1999)
Vapnik, V.N.: An overview of statistical learning theory. IEEE transactions on neural networks 10(5), 988–999 (1999)
work page 1999
-
[8]
Cambridge university press, Cambridge, UK (2014)
Shalev-Shwartz, S., Ben-David, S.: Understanding Machine Learning: From Theory to Algorithms. Cambridge university press, Cambridge, UK (2014)
work page 2014
-
[9]
Desislavov, R., Mart´ ınez-Plumed, F., Hern´ andez-Orallo, J.: Compute and energy consumption trends in deep learning inference. arXiv:2109.05472 (2021)
-
[10]
Computer Methods and Programs in Biomedicine 213, 106504 (2022)
Bailly, A., Blanc, C., Francis, ´E., Guillotin, T., Jamal, F., Wakim, B., Roy, P.: Effects of dataset size and interactions on the prediction performance of logis- tic regression and deep learning models. Computer Methods and Programs in Biomedicine 213, 106504 (2022)
work page 2022
-
[11]
In: Proceedings of the IEEE International Conference on Computer Vision, pp
Sun, C., Shrivastava, A., Singh, S., Gupta, A.: Revisiting unreasonable effec- tiveness of data in deep learning era. In: Proceedings of the IEEE International Conference on Computer Vision, pp. 843–852 (2017)
work page 2017
-
[12]
Artificial Intelligence Review (2023)
Meyers, C., L¨ ofstedt, T., Elmroth, E.: Safety-critical computer vision: An empiri- cal survey of adversarial evasion attacks and defenses on computer vision systems. Artificial Intelligence Review (2023)
work page 2023
-
[13]
O’Brien, M., Fingerhut, H., Press, T.A.: A.I. tools fueled a 34% spike in Microsoft’s water consumption, and one city with its data centers is concerned about the effect on residential supply. Fortune (2023). https://fortune.com/2023/ 24 09/09/ai-chatgpt-usage-fuels-spike-in-microsoft-water-consumption/
work page 2023
-
[14]
The History Began from AlexNet: A Comprehensive Survey on Deep Learning Approaches
Alom, M.Z., Taha, T.M., Yakopcic, C., Westberg, S., Sidike, P., Nasrin, M.S., Van Esesn, B.C., Awwal, A.A.S., Asari, V.K.: The history began from alexnet: A comprehensive survey on deep learning approaches. arXiv preprint arXiv:1803.01164 (2018)
work page internal anchor Pith review Pith/arXiv arXiv 2018
-
[15]
In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp
He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recogni- tion. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770–778 (2016)
work page 2016
-
[16]
An Empirical Study of Mamba-based Language Models
Waleffe, R., Byeon, W., Riach, D., Norick, B., Korthikanti, V., Dao, T., Gu, A., Hatamizadeh, A., Singh, S., Narayanan, D., et al.: An empirical study of mamba-based language models. arXiv preprint arXiv:2406.07887 (2024)
work page internal anchor Pith review Pith/arXiv arXiv 2024
-
[17]
International Elec- trotechnical Commission (2010)
Commission, I.E.: IEC 61508 Safety and Functional Safety. International Elec- trotechnical Commission (2010)
work page 2010
-
[18]
https://www.iso.org/standard/43464.html (2018)
International Standards Organization: ISO 26262-1:2011, Road vehicles — Func- tional safety. https://www.iso.org/standard/43464.html (2018)
work page 2011
-
[19]
In: 2011 IEEE Long Island Systems, Applications and Technology Conference, pp
Axelrod, C.W.: Applying lessons from safety-critical systems to security-critical software. In: 2011 IEEE Long Island Systems, Applications and Technology Conference, pp. 1–6 (2011). IEEE
work page 2011
-
[20]
Safety Science 77, 133–142 (2015)
Acharyulu, P.S., Seetharamaiah, P.: A framework for safety automation of safety- critical systems operations. Safety Science 77, 133–142 (2015)
work page 2015
-
[21]
Brown, T.B., Man´ e, D., Roy, A., Abadi, M., Gilmer, J.: Adversarial patch. arXiv:1712.09665 (2017)
work page internal anchor Pith review Pith/arXiv arXiv 2017
-
[22]
In: 2017 IEEE Symposium on Security and Privacy (SP), pp
Carlini, N., Wagner, D.: Towards evaluating the robustness of neural networks. In: 2017 IEEE Symposium on Security and Privacy (SP), pp. 39–57 (2017)
work page 2017
-
[23]
In: International Conference on Machine Learning, pp
Croce, F., Hein, M.: Reliable evaluation of adversarial robustness with an ensem- ble of diverse parameter-free attacks. In: International Conference on Machine Learning, pp. 2206–2216 (2020). PMLR
work page 2020
-
[24]
In: IEEE Symposium on Security and Privacy (SP), pp
Chen, J., Jordan, M.I., Wainwright, M.J.: HopSkipJumpAttack: A query-efficient decision-based attack. In: IEEE Symposium on Security and Privacy (SP), pp. 1277–1294 (2020). IEEE
work page 2020
-
[26]
Nicolae, M.I., Sinn, M., Tran, M.N., Buesser, B., Rawat, A., Wistuba, M., Zantedeschi, V., Baracaldo, N., Chen, B., Ludwig, H., Molloy, I., Edwards, B.: Adversarial robustness toolbox v1.2.0. CoRR 1807.01069 (2018) 25
-
[27]
arXiv preprint arXiv:2403.00787 (2024)
Panchal, D., Verma, P., Baran, I., Musgrove, D., Lu, D.: Reusable mlops: Reusable deployment, reusable infrastructure and hot-swappable machine learning models and services. arXiv preprint arXiv:2403.00787 (2024)
-
[28]
In: 2017 IEEE International Conference on Software Architecture Workshops (ICSAW), pp
Hasselbring, W., Steinacker, G.: Microservice architectures for scalability, agility and reliability in e-commerce. In: 2017 IEEE International Conference on Software Architecture Workshops (ICSAW), pp. 243–246 (2017). IEEE
work page 2017
-
[29]
IEEE Transactions on Cloud Computing (2022)
Zhou, R., Pang, J., Zhang, Q., Wu, C., Jiao, L., Zhong, Y., Li, Z.: Online scheduling algorithm for heterogeneous distributed machine learning jobs. IEEE Transactions on Cloud Computing (2022)
work page 2022
-
[30]
Journal of Cloud Computing 12(1), 4 (2023)
Singh, N., Hamid, Y., Juneja, S., Srivastava, G., Dhiman, G., Gadekallu, T.R., Shah, M.A.: Load balancing and service discovery using docker swarm for microservice based big data applications. Journal of Cloud Computing 12(1), 4 (2023)
work page 2023
-
[31]
Github: Octoverse Projects. Github.com (2019). https://octoverse.github.com/ 2018/projects.html
work page 2019
-
[32]
arXiv preprint arXiv:1907.10701 (2019)
Wang, Y.E., Wei, G.-Y., Brooks, D.: Benchmarking tpu, gpu, and cpu platforms for deep learning. arXiv preprint arXiv:1907.10701 (2019)
-
[33]
Kubernetes: Kubernetes–an open source system for managing containerized applications. Github (2019). https://github.com/kubernetes/kubernetes
work page 2019
-
[34]
dvc.org: DVC- Data Version Control. Github (2023). https://github.com/ iterative/dvc.org
work page 2023
-
[35]
Yadan, O.: Hydra - A framework for elegantly configuring complex applications. Github (2019). https://github.com/facebookresearch/hydra
work page 2019
-
[36]
Advances in neural information processing systems 32 (2019)
Cao, Y., Gu, Q.: Generalization bounds of stochastic gradient descent for wide and deep neural networks. Advances in neural information processing systems 32 (2019)
work page 2019
-
[37]
Journal of Machine Learning Research 23(173), 1–65 (2022)
Granziol, D., Zohren, S., Roberts, S.: Learning rates as a function of batch size: A random matrix theory approach to neural network training. Journal of Machine Learning Research 23(173), 1–65 (2022)
work page 2022
-
[38]
In: Artificial Intelligence and Machine Learning for Multi-domain Operations Applications, vol
Smith, L.N., Topin, N.: Super-convergence: Very fast training of neural networks using large learning rates. In: Artificial Intelligence and Machine Learning for Multi-domain Operations Applications, vol. 11006, pp. 369–386 (2019). SPIE
work page 2019
-
[39]
In: International Conference on Cloud, Big Data and Trust, vol
Sajid, M., Raza, Z.: Cloud computing: Issues & challenges. In: International Conference on Cloud, Big Data and Trust, vol. 20, pp. 13–15 (2013)
work page 2013
-
[40]
Poisoning Attacks against Support Vector Machines
Biggio, B., Nelson, B., Laskov, P.: Poisoning Attacks against Support Vector 26 Machines. arXiv:1206.6389 [cs, stat] (2013)
work page internal anchor Pith review Pith/arXiv arXiv 2013
-
[41]
In: Proceedings of the AAAI Conference on Artificial Intelligence, vol
Saha, A., Subramanya, A., Pirsiavash, H.: Hidden trigger backdoor attacks. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 34, pp. 11957– 11965 (2020)
work page 2020
-
[42]
Adversarial Attacks and Defences: A Survey
Chakraborty, A., Alam, M., Dey, V., Chattopadhyay, A., Mukhopadhyay, D.: Adversarial attacks and defences: A survey. arXiv:1810.00069 [cs, stat] (2018)
work page internal anchor Pith review Pith/arXiv arXiv 2018
-
[43]
In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp
Orekondy, T., Schiele, B., Fritz, M.: Knockoff nets: Stealing functionality of black- box models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4954–4963 (2019)
work page 2019
-
[44]
In: International Conference on Machine Learning, pp
Choquette-Choo, C.A., Tramer, F., Carlini, N., Papernot, N.: Label-only mem- bership inference attacks. In: International Conference on Machine Learning, pp. 1964–1974 (2021). PMLR
work page 1964
-
[45]
In: Proceedings of the 2021 ACM SIGSAC Conference on Computer and Communications Security, pp
Li, Z., Zhang, Y.: Membership leakage in label-only exposures. In: Proceedings of the 2021 ACM SIGSAC Conference on Computer and Communications Security, pp. 880–895 (2021)
work page 2021
-
[46]
PloS one 17(4), 0265723 (2022)
Kotyan, S., Vargas, D.V.: Adversarial robustness assessment. PloS one 17(4), 0265723 (2022)
work page 2022
-
[47]
Explaining and Harnessing Adversarial Examples
Goodfellow, I.J., Shlens, J., Szegedy, C.: Explaining and harnessing adversarial examples. arXiv:1412.6572 (2014)
work page internal anchor Pith review Pith/arXiv arXiv 2014
-
[48]
In: 29th USENIX Security Symposium (USENIX Security 20), pp
Leurent, G., Peyrin, T.: SHA-1 is a shambles: First Chosen-Prefix collision on SHA-1 and application to the PGP web of trust. In: 29th USENIX Security Symposium (USENIX Security 20), pp. 1839–1856 (2020)
work page 2020
-
[49]
Kamal, P.: A study on the security of password hashing based on gpu based, password cracking using high-performance cloud computing. Master’s thesis, St. Cloud State. St. Cloud, Minnesota, USA. (2017)
work page 2017
-
[50]
Biggio, B., Corona, I., Maiorca, D., Nelson, B., ˇSrndi´ c, N., Laskov, P., Giacinto, G., Roli, F.: Evasion attacks against machine learning at test time. In: Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2013, Prague, Czech Republic, September 23-27, 2013, Proceedings, Part III 13, pp. 387–402 (2013). Springer
work page 2013
-
[51]
British journal of cancer 89(3), 431–436 (2003)
Bradburn, M.J., Clark, T.G., Love, S.B., Altman, D.G.: Survival analysis part II: multivariate data analysis–an introduction to concepts and methods. British journal of cancer 89(3), 431–436 (2003)
work page 2003
-
[52]
Springer, New York, NY, USA (2012) 27
Kleinbaum, D.G., Klein, M.: Survival Analysis a Self-learning Text. Springer, New York, NY, USA (2012) 27
work page 2012
-
[53]
arXiv preprint arXiv:2106.04525 (2021)
Tan, J., Yang, J., Wu, S., Chen, G., Zhao, J.: A critical look at the current train/test split in machine learning. arXiv preprint arXiv:2106.04525 (2021)
-
[54]
In: Proceedings of the 36th International Conference on Machine Learning
Dohmatob, E.: Generalized No Free Lunch Theorem for Adversarial Robust- ness. In: Proceedings of the 36th International Conference on Machine Learning. PMLR, vol. 97 (2019)
work page 2019
-
[55]
In: Accepted for International Conference on Machine Learning and Cybernetics (2023)
Meyers, C., Reza, M., L¨ ofstedt, T., Elmroth, E.: A systematic approach to robust- ness modelling. In: Accepted for International Conference on Machine Learning and Cybernetics (2023)
work page 2023
-
[56]
In: Pro- ceedings of the 2020 Genetic and Evolutionary Computation Conference, pp
Ozaki, Y., Tanigaki, Y., Watanabe, S., Onishi, M.: Multiobjective tree-structured parzen estimator for computationally expensive optimization problems. In: Pro- ceedings of the 2020 Genetic and Evolutionary Computation Conference, pp. 533–541 (2020)
work page 2020
-
[57]
Akiba, T., Sano, S., Yanase, T., Ohta, T., Koyama, M.: Optuna: A next- generation hyperparameter optimization framework. In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp. 2623–2631 (2019)
work page 2019
-
[58]
arXiv preprint arXiv:2304.11127 (2023)
Watanabe, S.: Tree-structured parzen estimator: Understanding its algorithm components and their roles for better empirical performance. arXiv preprint arXiv:2304.11127 (2023)
-
[59]
Statistics in Medicine 39(21), 2714–2742 (2020)
Austin, P.C., Harrell Jr, F.E., Klaveren, D.: Graphical calibration curves and the integrated calibration index (ICI) for survival models. Statistics in Medicine 39(21), 2714–2742 (2020)
work page 2020
-
[60]
Very Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014)
work page internal anchor Pith review Pith/arXiv arXiv 2014
-
[61]
In: 2023 IEEE 16th International Conference on Cloud Computing (CLOUD), pp
Amaral, M., Chen, H., Chiba, T., Nakazawa, R., Choochotkaew, S., Lee, E.K., Eilam, T.: Kepler: A framework to calculate the energy consumption of con- tainerized applications. In: 2023 IEEE 16th International Conference on Cloud Computing (CLOUD), pp. 69–71 (2023)
work page 2023
-
[62]
In: 2022 IEEE International Conference on Service- Oriented System Engineering (SOSE), pp
Sedghpour, M.R.S., Townend, P.: Service mesh and ebpf-powered microservices: A survey and future directions. In: 2022 IEEE International Conference on Service- Oriented System Engineering (SOSE), pp. 176–184 (2022). https://doi.org/10. 1109/SOSE55356.2022.00027
-
[63]
Multiobjective optimization: Interactive and evolutionary approaches, 373–404 (2008)
Zitzler, E., Knowles, J., Thiele, L.: Quality assessment of pareto set approxi- mations. Multiobjective optimization: Interactive and evolutionary approaches, 373–404 (2008)
work page 2008
-
[64]
Legriel, J., Le Guernic, C., Cotton, S., Maler, O.: Approximating the pareto front 28 of multi-criteria optimization problems. In: International Conference on Tools and Algorithms for the Construction and Analysis of Systems, pp. 69–83 (2010). Springer
work page 2010
-
[65]
IEEE Signal Processing Magazine 29(6), 141–142 (2012)
Deng, L.: The MNIST database of handwritten digit images for machine learning research. IEEE Signal Processing Magazine 29(6), 141–142 (2012)
work page 2012
-
[66]
Technical report, Toronto, ON, Canada (2009)
Krizhevsky, A., Hinton, G., et al.: Learning multiple layers of features from tiny images. Technical report, Toronto, ON, Canada (2009)
work page 2009
-
[67]
Towards Deep Learning Models Resistant to Adversarial Attacks
Madry, A., Makelov, A., Schmidt, L., Tsipras, D., Vladu, A.: Towards deep learning models resistant to adversarial attacks. arXiv:1706.06083 (2017)
work page internal anchor Pith review Pith/arXiv arXiv 2017
-
[68]
In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp
Moosavi-Dezfooli, S.-M., Fawzi, A., Frossard, P.: Deepfool: a simple and accurate method to fool deep neural networks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 2574–2582 (2016)
work page 2016
-
[69]
Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networks
Xu, W., Evans, D., Qi, Y.: Feature squeezing: Detecting adversarial examples in deep neural networks. arXiv:1704.01155 (2017)
work page internal anchor Pith review Pith/arXiv arXiv 2017
-
[70]
Svedin, M., Chien, S.W., Chikafa, G., Jansson, N., Podobas, A.: Benchmarking the nvidia gpu lineage: From early k80 to modern a100 with asynchronous memory transfers. In: Proceedings of the 11th International Symposium on Highly Efficient Accelerators and Reconfigurable Technologies, pp. 1–6 (2021)
work page 2021
-
[71]
Xu, R., Han, F., Ta, Q.: Deep learning at scale on nvidia v100 accelerators. In: 2018 IEEE/ACM Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS), pp. 23–32 (2018). IEEE
work page 2018
-
[72]
Compute Engine: Virtual Machines (VMS)
Google: GPU pricing. Compute Engine: Virtual Machines (VMS). google cloud. Google. https://cloud.google.com/compute/gpus-pricing
-
[73]
Compute Engine: Virtual Machines (VMS)
Google: GPU pricing. Compute Engine: Virtual Machines (VMS). google cloud. Google. https://cloud.google.com/compute/vm-instance-pricing# accelerator-optimised
-
[74]
lifelines: survival analysis in Python.Journal of Open Source Software, 4(40):1317, 2019
Davidson-Pilon, C.: lifelines: survival analysis in python. Journal of Open Source Software 4(40), 1317 (2019) https://doi.org/10.21105/joss.01317
-
[75]
IEEE Communications surveys & tutorials 18(1), 732–794 (2015)
Dayarathna, M., Wen, Y., Fan, R.: Data center energy consumption modeling: A survey. IEEE Communications surveys & tutorials 18(1), 732–794 (2015)
work page 2015
-
[76]
Statistics in medicine 19(4), 441–452 (2000)
Schmoor, C., Sauerbrei, W., Schumacher, M.: Sample size considerations for the evaluation of prognostic factors in survival analysis. Statistics in medicine 19(4), 441–452 (2000)
work page 2000
-
[77]
Controlled clinical trials 2(2), 93–113 (1981) 29
Lachin, J.M.: Introduction to sample size determination and power analysis for clinical trials. Controlled clinical trials 2(2), 93–113 (1981) 29
work page 1981
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.