Pith. sign in

REVIEW 4 major objections 4 minor 44 references

Continual Learning Strategies for 3D Engineering Regression Problems: A Benchmarking Study

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

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

Pith's one-line read A benchmarking study claims that a simple replay buffer lets engineering surrogate models learn from new data nearly as well as retraining from scratch, at about half the compute.

desk verdict Solid first benchmark of continual learning on engineering regression surrogates; Replay's near-Joint accuracy is plausible and useful, but missing evaluation details and a handicapped GEM make the headline result conditional. read the letter →

arxiv 2504.12503 v1 pith:CVBOM46G submitted 2025-04-16 cs.LG cs.AIcs.CE

Kaira M. Samuel , Faez Ahmed This is my paper · ORCID
classification cs.LGcs.AIcs.CE
keywords continuallearningcatastrophicforgettingengineeringregressionsurrogatemodelingexperiencereplaybenchmarkingpointcloudsdragcoefficientprediction
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

Engineers often train surrogate models—neural networks that predict continuous quantities such as drag coefficients from 3D geometry—on static datasets, but real design pipelines add new data over time, making retraining from scratch expensive. This paper transfers continual learning, the study of training models on sequential data without catastrophic forgetting, to engineering regression by constructing nine benchmarks from five engineering datasets. It proposes three regression-specific task sequences: bin-incremental (target values partitioned into ranges), input-incremental (input categories or operating conditions), and multi-target-incremental (new output quantities). Benchmarking Experience Replay, Elastic Weight Consolidation, and Gradient Episodic Memory, the paper claims that Replay, using only 20% of past data as a memory buffer, reaches performance comparable to fully retraining on several benchmarks while reducing training time by nearly half. A sympathetic reader would take the central message to be that simple rehearsal, not parameter regularization, is the practical route to continual surrogate models.

What carries the argument

The central device is the pairing of regression-specific task sequences with a rehearsal buffer. In the bin-incremental scenario the continuous target is cut into ordered value ranges and each range becomes one experience; in the input-incremental scenario experiences are defined by input categories such as car type; a third multi-target scenario, noted but not benchmarked, would add new output quantities. Experience Replay carries the argument by storing a random 20% of past samples in a memory buffer and mixing them into training on each new experience, which is what lets the model retain old mappings. EWC's Fisher-information penalty and GEM's gradient-projection constraint serve as the comparators that Replay beats. The forgetting ratio, a normalized version of the maximum regression-error increase on earlier experiences, is the metric that makes the cross-benchmark ranking possible.

What would settle it

Run Experience Replay and EWC on a new engineering regression stream where target bins overlap and tasks are defined by operating conditions rather than value ranges, evaluating on a held-out test set excluded from every replay buffer; if Replay's final error does not come close to Joint retraining, or if EWC matches Replay, the paper's central advantage fails.

Watch

Extended reading notes

Core claim

On its nine benchmarks, the paper claims a clear ranking: Experience Replay—randomly replaying 20% of prior samples—is the best continual learning strategy, with final mean percent error close to the Joint retraining baseline (for instance 2.07% vs. 1.57% on DrivAerNet point clouds and 3.64% vs. 3.36% on DrivAerNet++ point clouds). GEM, which constrains gradients using stored past patterns, ranks between Replay and the regularization method EWC, and on forgetting Replay and GEM are statistically indistinguishable and sit just below Joint. EWC performs no better than naive sequential training on most benchmarks. The input-incremental scenario (car category as task) produces less forgetting overall than the bin-incremental scenario (drag-value range as task), and on the small glider dataset all methods show negative forgetting, meaning learning later experiences improved predictions on earlier ones.

Load-bearing premise

The conclusions assume the benchmark scenarios—splitting data into disjoint target-value bins or by car category—mirror how engineering data actually arrives over time; if real streams have overlapping target ranges or different task boundaries, the measured ranking of strategies may not transfer.

Editorial extensions

If this is right

  • Engineers can update surrogate models on streaming design data with a 20% replay buffer and expect final prediction error close to what full retraining would give, at roughly half the training time.
  • Rehearsal-based strategies (Replay and GEM) are statistically indistinguishable on forgetting and rank just below the ideal joint-retraining baseline, so the choice between them can be made on compute grounds.
  • Elastic Weight Consolidation does not beat naive sequential training on most of these regression benchmarks, suggesting regularization penalties are not enough when targets are continuous.
  • In the input-incremental car-category scenario, forgetting is less severe than in the bin-incremental scenario, and models recover after the second category, implying shared geometry between tasks eases retention.
  • On small datasets with narrow target ranges, learning later experiences can improve predictions on earlier ones (negative forgetting), so catastrophic forgetting is not automatic in engineering regression.

Reading between the lines

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

  • The paper's compute comparison counts total runtime including evaluation; the memory cost of a 20% replay buffer is not tabulated. If buffer memory is counted, replay's advantage over memory-free EWC could narrow in settings where storage is the bottleneck.
  • A testable extension follows from the input-incremental results: because the second car category caused an error spike followed by recovery, replay methods may act as a cheap concept-drift response whenever shifting operating conditions periodically revisit previously seen regimes—something the paper does not claim.
  • If replay at 20% buffer generalizes beyond these nine benchmarks, engineering MLOps pipelines could replace full retraining with rolling buffers, freeing compute for simulation; that practical workflow implication is implicit, not proven.
  • The proposed multi-target-incremental scenario (adding new output quantities over time) is not benchmarked, so the paper's ranking of strategies does not extend to that regime; testing it is a direct next step.
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 / 4 minor

Summary. The paper introduces continual learning (CL) to 3D engineering regression problems by proposing three regression-specific CL scenarios (bin-incremental, input-incremental, and multi-target incremental) and constructing nine benchmarks from five engineering datasets (SHIPD, DrivAerNet, DrivAerNet++, ShapeNet Car, RAADL). It benchmarks three standard CL strategies from the Avalanche library—Experience Replay (ER), Elastic Weight Consolidation (EWC), and Gradient Episodic Memory (GEM)—against naive incremental training and a Joint retraining baseline, reporting mean percent error, mean absolute error, forgetting ratios, and total runtime. The central empirical claim is that Replay achieves performance comparable to Joint retraining on several benchmarks while reducing training time by roughly half, whereas EWC and GEM are less effective.

Significance. If the central claim holds, the paper provides practically valuable evidence that a simple replay buffer of 20% of past data can nearly match joint retraining for evolving engineering surrogate models at substantially reduced compute. The proposed benchmarks and scenarios fill a clear gap, since most CL research targets classification rather than regression, and engineering domains have been largely overlooked. The use of standard Avalanche implementations and an independent Joint baseline are strengths that increase the credibility of the comparison. However, the evaluation protocol—especially the construction of test sets relative to experience bins—is not described, and several hyperparameters are unreported. These gaps currently prevent the reader from verifying whether the headline result is an artifact of leakage or unfair configuration, so the contribution is not yet fully substantiated.

major comments (4)
  1. [Section 5 (Implementation and Evaluation Metrics), Tables 3–12] The manuscript never specifies how the test sets were constructed relative to the experience bins. It is not stated whether each experience's test samples are disjoint from the training and replay data, how the global train/test partition was made, or whether duplicate or near-duplicate geometries across experiences were removed. This is load-bearing for the central claim: if the replay buffer is sampled from the same pool that is later scored, Replay's reported MAE and forgetting advantages over EWC and GEM could be inflated. Please describe the exact train/test split procedure, confirm that replay and test sets do not overlap, and provide the number of samples in each split.
  2. [Table 13 and Section 5 (Implementation and Evaluation Metrics)] Section 5 states that all methods were constrained to have a shorter total runtime than the Joint retraining baseline, but Table 13 lists GEM at 8168.27 s on SplitDRIVAERNET++-PC (Bin) versus Joint at 7662.63 s, violating the constraint. Additionally, Section 7 concedes that GEM's patterns-per-experience had to be reduced to meet this cap, which 'likely contributed to GEM's diminished overall performance.' Thus the runtime constraint is not consistently enforced, and the comparison of GEM against Replay and Joint is confounded by an unequal compute budget. Please either re-run GEM without the artificial pattern reduction, or explicitly relax the constraint and discuss the implications for the ranking.
  3. [Section 5 (Surrogate Models; Continual Learning Strategies)] Hyperparameters for the continual learning strategies are not reported: the EWC regularization strength lambda, GEM's patterns per experience (PPE), the replay buffer size, learning rate, number of epochs, and batch size are all absent. Section 5 states that the surrogate models were tuned 'using a hit-and-trial approach,' but no values or search ranges are given. Since the paper's headline conclusion is a ranking of strategies, the validity of that ranking depends on all methods being reasonably configured; unreported or arbitrarily chosen hyperparameters can change the ordering (e.g., a poorly chosen EWC lambda could explain EWC's near-Naive performance). Please report the exact hyperparameter values, the tuning procedure, and ideally perform a sensitivity analysis for the key parameters.
  4. [Section 7 (Statistical Trends Across Benchmarks)] The Mann-Whitney U test analysis and critical-difference plots are described only briefly. It is unclear what the unit of analysis is (nine benchmarks, or multiple seeds per benchmark), how the metrics were normalized before ranking, and whether any multiple-comparison correction was applied. The claim that Replay and GEM are 'statistically indistinguishable' on forgetting, while Replay is significantly better on final MAE, needs these details to be assessable. Please specify the exact statistical procedure, including the number of independent measurements per strategy, the normalization formula, and the significance level used.
minor comments (4)
  1. [Section 5 (Implementation and Evaluation Metrics)] The definition of 'best forgetting ratio' as 'the lowest FRk achieved over all trials' is unclear; please specify the number of trials/seeds and justify why the best rather than the mean (or median) is reported. Reporting the mean with variance would be more standard.
  2. [Section 4 (Benchmarks)] There is a typo: 'illustrted' should be 'illustrated' in the sentence describing the benchmarks in Figure 4.
  3. [References] Some references are incomplete: [16] and [17] (Rolnick et al.; Lopez-Paz and Ranzato) lack year and venue information, and [42] is listed as 'n.d.' with only a URL. Please provide full bibliographic details for all entries.
  4. [Figure 6] The subplots in Figure 6 are small and the axis labels appear difficult to read at publication size. Consider enlarging the figure or splitting it into separate panels.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity found: benchmark results are measured against an independent Joint retraining baseline; author-group datasets are inputs, not load-bearing citations.

full rationale

The paper's central claim (Replay achieves performance comparable to retraining at roughly half the training time) is an empirical benchmark result, not a derivation. The comparison is made against a Joint retraining baseline computed on the same data, which is an independent yardstick: Replay's MPE, MAE, and forgetting ratios are measured errors, not quantities defined in terms of the baseline or of any fitted parameter. The CL strategies are standard Avalanche implementations (Experience Replay, EWC, GEM) with stated hyperparameters and a 20% replay-buffer constraint, so no fitted input is renamed as a prediction. The proposed regression scenarios (bin incremental, input incremental, multi-target incremental) are explicit definitions for splitting data into experiences; no claim is derived from them by construction. The only author-overlap dependence is dataset provenance: SHIPD, DrivAerNet, DrivAerNet++, and the ShapeNet Car drag surrogate come from the same research group, but the benchmark conclusions do not reduce to claims in those papers—the datasets are inputs and the Joint baseline provides an external yardstick. Section 7's statement that 'These conclusions are based on the assumption that our benchmark suite... offers representative coverage of 3D engineering problems' is a limitation statement, not circularity. Two evaluation-protocol concerns—the paper never specifies how test sets were split relative to experience bins, and Table 13 lists GEM at 8168.27 s on SplitDRIVAERNET++-PC (Bin) versus Joint at 7662.63 s, contradicting the stated runtime constraint—are correctness risks, not circular reductions. No uniqueness theorem is imported, no ansatz is smuggled via citation, and no known result is merely renamed. The score of 2 reflects the multiple non-load-bearing self-citations for datasets, which do not make the central comparison circular.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The paper's quantitative results depend on five sets of numbers the authors chose: the replay buffer size (20% of data), the EWC regularization strength (unreported), the GEM patterns per experience (unreported and explicitly reduced to satisfy a runtime constraint), the number of experiences per benchmark (4 bin, 3 input), and the surrogate model hyperparameters (selected by hit-and-trial, unreported). No physically or mathematically fixed constants are involved. The scenarios rest on domain assumptions about how engineering data arrives, and the fidelity of the Avalanche implementations and the pretrained PointNet are taken as given.

free parameters (5)
  • Replay buffer size = 20% of dataset
    Set exogenously as a storage constraint (Section 5), not fitted; affects all replay results.
  • EWC regularization strength lambda = not reported
    Controls the Fisher penalty in L_EWC = L_new + lambda/2 * sum F_i (theta_i - theta*_i)^2 (Section 5); authors tuned it by hit-and-trial but do not report the value.
  • GEM patterns per experience (PPE) = not reported
    Authors explicitly restricted PPE to keep runtime below Joint (Section 7), which likely weakened GEM; exact count not given.
  • Number of experiences/bins = 4 for bin-incremental, 3 for input-incremental
    Bin boundaries and counts chosen by authors; benchmark-construction choice used throughout Tables 3-12.
  • Surrogate model hyperparameters (learning rate, epochs, batch size) = not reported
    Selected by 'hit-and-trial' on a validation set (Section 5); final values not listed.
assumptions (5)
  • domain assumption The five datasets are representative of 3D engineering regression problems
    Used to generalize the benchmark conclusions beyond the specific datasets (Section 7).
  • domain assumption Experience boundaries in bin/input incremental scenarios match real engineering data flow
    The scenarios are proposed as realistic (Section 4); conclusions depend on this representativeness.
  • domain assumption Avalanche's implementations of ER, EWC, and GEM are faithful to the published methods
    All experiments use Avalanche (Section 5); no independent verification is provided.
  • domain assumption Pretrained PointNet features transfer from ScanObjectNN to engineering point clouds
    To fight overfitting, a PointNet pretrained on indoor scans is frozen except outer layers (Section 5); if transfer is poor, all point-cloud results are affected.
  • standard math Mann-Whitney U test assumptions hold for normalized benchmark metrics
    Used for critical difference plots in Section 7.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Continual Learning Strategies for 3D Engineering Regression Problems: A Benchmarking Study." pith.science (2026). https://pith.science/paper/CVBOM46G

@misc{pith2026250412503,
  author       = {Pith},
  title        = {Pith review of: Continual Learning Strategies for 3D Engineering Regression Problems: A Benchmarking Study},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CVBOM46G}},
  note         = {Machine review of arXiv:2504.12503}
}
read the original abstract

Engineering problems that apply machine learning often involve computationally intensive methods but rely on limited datasets. As engineering data evolves with new designs and constraints, models must incorporate new knowledge over time. However, high computational costs make retraining models from scratch infeasible. Continual learning (CL) offers a promising solution by enabling models to learn from sequential data while mitigating catastrophic forgetting, where a model forgets previously learned mappings. This work introduces CL to engineering design by benchmarking several CL methods on representative regression tasks. We apply these strategies to five engineering datasets and construct nine new engineering CL benchmarks to evaluate their ability to address forgetting and improve generalization. Preliminary results show that applying existing CL methods to these tasks improves performance over naive baselines. In particular, the Replay strategy achieved performance comparable to retraining in several benchmarks while reducing training time by nearly half, demonstrating its potential for real-world engineering workflows. The code and datasets used in this work will be available at: https://github.com/kmsamuel/cl-for-engineering-release.

Figures

Figures reproduced from arXiv: 2504.12503 by the authors.

Figure 1
Figure 1. Catastrophic forgetting is demonstrated using the car drag prediction problem from the DrivAerNet dataset. A [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Continual learning model training scheme, in which a model is incrementally trained on new experiences and evalu [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Proposed regression continual learning scenarios that would be relevant for engineering. Figure [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Continual learning benchmarks are defined as the combination of a dataset with a continual learning scenario. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Overview of the continual learning strategies evaluated in this work. In Figure [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Incremental MAE and Forgetting across all benchmarks, including both bin-incremental and input-incremental (in [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: Rank plots of CL strategies for Final MAE, Forgetting Ratio, and Total Run Time across all nine benchmarks [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 41 canonical work pages

  1. [1]

    A review of the artifi- cial neural network surrogate modeling in aerodynamic design,

    Sun, G., and Wang, S., 2019, “A review of the artifi- cial neural network surrogate modeling in aerodynamic design,” Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering, 233(16), pp. 5863–5872

    work page 2019

  2. [2]

    Drivaer- net: A parametric car dataset for data-driven aerody- namic design and prediction,

    Elrefaie, M., Dai, A., and Ahmed, F., 2025, “Drivaer- net: A parametric car dataset for data-driven aerody- namic design and prediction,” Journal of Mechanical Design, 147(4)

    work page 2025

  3. [3]

    Surrogate modeling of car drag coefficient with depth and normal renderings,

    Song, B., Yuan, C., Permenter, F., Arechiga, N., and Ahmed, F., 2023, “Surrogate modeling of car drag coefficient with depth and normal renderings,” In In- ternational Design Engineering Technical Conferences and Computers and Information in Engineering Con- ference, V ol. 87301, American Society of Mechanical Engineers, p. V03AT03A029

    work page 2023

  4. [4]

    Fast predictions of aircraft aerodynamics using deep- learning techniques,

    Sabater, C., St ¨urmer, P., and Bekemeyer, P., 2022, “Fast predictions of aircraft aerodynamics using deep- learning techniques,” AIAA Journal, 60(9), pp. 5249– 5261

    work page 2022

  5. [5]

    Data-driven design for metamaterials and multiscale systems: a review,

    Lee, D., Chen, W., Wang, L., Chan, Y .-C., and Chen, W., 2024, “Data-driven design for metamaterials and multiscale systems: a review,” Advanced Materials, 36(8), p. 2305254

    work page 2024

  6. [6]

    Digital twins: state-of-the-art and future direc- tions for modeling and simulation in engineering dy- namics applications,

    Wagg, D., Worden, K., Barthorpe, R., and Gardner, P., 2020, “Digital twins: state-of-the-art and future direc- tions for modeling and simulation in engineering dy- namics applications,” ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part B: Mechani- cal Engineering, 6(3), p. 030901

    work page 2020

  7. [7]

    Industrial applications of digital twins,

    Jiang, Y ., Yin, S., Li, K., Luo, H., and Kaynak, O., 2021, “Industrial applications of digital twins,” Philosophical Transactions of the Royal Society A, 379(2207), p. 20200360

    work page 2021

  8. [8]

    Continual learning: Applications and the road forward,

    Verwimp, E., Aljundi, R., Ben-David, S., Bethge, M., Cossu, A., Gepperth, A., Hayes, T. L., H¨ullermeier, E., Kanan, C., Kudithipudi, D., et al., 2023, “Continual learning: Applications and the road forward,” arXiv preprint arXiv:2311.11908

    arXiv 2023

Show all 44 references
  1. [9]

    A comprehensive study of class incremental learning algorithms for visual tasks,

    Belouadah, E., Popescu, A., and Kanellos, I., 2021, “A comprehensive study of class incremental learning algorithms for visual tasks,” Neural Networks, 135, pp. 38–54

    2021

  2. [10]

    Class- incremental learning: survey and performance eval- uation on image classification,

    Masana, M., Liu, X., Twardowski, B., Menta, M., Bag- danov, A. D., and Van De Weijer, J., 2022, “Class- incremental learning: survey and performance eval- uation on image classification,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(5), pp. 5513–5533

    2022

  3. [11]

    Gradient-based learning applied to document recogni- tion,

    Lecun, Y ., Bottou, L., Bengio, Y ., and Haffner, P., 1998, “Gradient-based learning applied to document recogni- tion,” Proceedings of the IEEE,86(11), pp. 2278–2324

    1998

  4. [12]

    Core50: a new dataset and benchmark for continuous object recogni- tion,

    Lomonaco, V ., and Maltoni, D., 2017, “Core50: a new dataset and benchmark for continuous object recogni- tion,” In Proceedings of the 1st Annual Conference on Robot Learning, S. Levine, V . Vanhoucke, and K. Gold- berg, eds., V ol. 78 ofProceedings of Machine Learning Researc...

    2017

  5. [13]

    Clear: An adaptive contin- ual learning framework for regression tasks,

    He, Y ., and Sick, B., 2021, “Clear: An adaptive contin- ual learning framework for regression tasks,” AI Per- spectives, 3(1), p. 2

    2021

  6. [14]

    Three types of incremental learning,

    Van de Ven, G. M., Tuytelaars, T., and Tolias, A. S., 2022, “Three types of incremental learning,” Nature Machine Intelligence, 4(12), pp. 1185–1197

    2022

  7. [15]

    Overcoming catastrophic forgetting in neural net- works,

    Kirkpatrick, J., Pascanu, R., Rabinowitz, N., Veness, J., Desjardins, G., Rusu, A. A., Milan, K., Quan, J., Ramalho, T., Grabska-Barwinska, A., et al., 2017, “Overcoming catastrophic forgetting in neural net- works,” Proceedings of the national academy of sci- ences, 114(13), ...

    2017

  8. [16]

    P., and Wayne, G., 2019, Experience replay for continual learning

    Rolnick, D., Ahuja, A., Schwarz, J., Lillicrap, T. P., and Wayne, G., 2019, Experience replay for continual learning

    2019

  9. [17]

    Lopez-Paz, D., and Ranzato, M., 2022, Gradient episodic memory for continual learning

    2022

  10. [18]

    A comprehensive survey of continual learning: Theory, method and application,

    Wang, L., Zhang, X., Su, H., and Zhu, J., 2024, “A comprehensive survey of continual learning: Theory, method and application,”IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(8), pp. 5362– 5383

    2024

  11. [19]

    A continual learning survey: Defying forgetting in classification tasks,

    De Lange, M., Aljundi, R., Masana, M., Parisot, S., Jia, X., Leonardis, A., Slabaugh, G., and Tuytelaars, T., 2021, “A continual learning survey: Defying forgetting in classification tasks,” IEEE transactions on pattern analysis and machine intelligence, 44(7), pp. 3366– 3385

    2021

  12. [20]

    Embracing change: Continual learning in deep neural networks,

    Hadsell, R., Rao, D., Rusu, A. A., and Pascanu, R., 2020, “Embracing change: Continual learning in deep neural networks,” Trends in cognitive sciences, 24(12), pp. 1028–1040

    2020

  13. [21]

    Online continual learning in image classification: An empirical survey,

    Mai, Z., Li, R., Jeong, J., Quispe, D., Kim, H., and Sanner, S., 2022, “Online continual learning in image classification: An empirical survey,” Neurocomputing, 469, pp. 28–51

    2022

  14. [22]

    Con- tinual learning should move beyond incremental classi- fication,

    Mitchell, R., Alliegro, A., Camoriano, R., Carri ´on- Ojeda, D., Carta, A., Chalvatzaki, G., Churamani, N., D’Eramo, C., Hamidi, S., Hesse, R., et al., 2025, “Con- tinual learning should move beyond incremental classi- fication,” arXiv preprint arXiv:2502.11927

    2025 arXiv

  15. [23]

    The clear benchmark: Continual learning on real- world imagery,

    Lin, Z., Shi, J., Pathak, D., and Ramanan, D., 2021, “The clear benchmark: Continual learning on real- world imagery,” In Thirty-fifth conference on neural in- formation processing systems datasets and benchmarks track (round 2)

    2021

  16. [24]

    Remind your neural network to pre- vent catastrophic forgetting,

    Hayes, T. L., Kafle, K., Shrestha, R., Acharya, M., and Kanan, C., 2020, “Remind your neural network to pre- vent catastrophic forgetting,” In European conference on computer vision, Springer, pp. 466–483

    2020

  17. [25]

    Online continual learning with maximal interfered re- trieval,

    Aljundi, R., Belilovsky, E., Tuytelaars, T., Charlin, L., Caccia, M., Lin, M., and Page-Caccia, L., 2019, “Online continual learning with maximal interfered re- trieval,” V ol. 32

    2019

  18. [26]

    Dark experience for general con- 17 Copyright © by ASME tinual learning: a strong, simple baseline,

    Buzzega, P., Boschini, M., Porrello, A., Abati, D., and Calderara, S., 2020, “Dark experience for general con- 17 Copyright © by ASME tinual learning: a strong, simple baseline,” V ol. 33, pp. 15920–15930

    2020

  19. [27]

    Con- tinual learning through synaptic intelligence,

    Zenke, F., Poole, B., and Ganguli, S., 2017, “Con- tinual learning through synaptic intelligence,” In In- ternational conference on machine learning, PMLR, pp. 3987–3995

    2017

  20. [28]

    Memory aware synapses: Learning what (not) to forget,

    Aljundi, R., Babiloni, F., Elhoseiny, M., Rohrbach, M., and Tuytelaars, T., 2018, “Memory aware synapses: Learning what (not) to forget,” In Proceedings of the European conference on computer vision (ECCV), pp. 139–154

    2018

  21. [29]

    Learning without forget- ting,

    Li, Z., and Hoiem, D., 2017, “Learning without forget- ting,” V ol. 40, IEEE, pp. 2935–2947

    2017

  22. [30]

    icarl: Incremental classifier and repre- sentation learning,

    Rebuffi, S.-A., Kolesnikov, A., Sperl, G., and Lampert, C. H., 2017, “icarl: Incremental classifier and repre- sentation learning,” In Proceedings of the IEEE con- ference on Computer Vision and Pattern Recognition, pp. 2001–2010

    2017

  23. [31]

    Op- timal continual learning has perfect memory and is np- hard,

    Knoblauch, J., Husain, H., and Diethe, T., 2020, “Op- timal continual learning has perfect memory and is np- hard,” pp. 5327–5337

    2020

  24. [32]

    Continual learning for neural regression networks to cope with concept drift in industrial processes using convex optimisation,

    Grote-Ramm, W., Lanuschny, D., Lorenzen, F., Brito, M. O., and Sch ¨onig, F., 2023, “Continual learning for neural regression networks to cope with concept drift in industrial processes using convex optimisation,” En- gineering Applications of Artificial Intelligence, 120, p. 105927

    2023

  25. [33]

    Miracle 3d: Memory-efficient integrated robust approach for con- tinual learning on point clouds via shape model con- struction,

    Resani, H., and Nasihatkon, B., 2024, “Miracle 3d: Memory-efficient integrated robust approach for con- tinual learning on point clouds via shape model con- struction,” arXiv preprint arXiv:2410.06418

    2024 arXiv

  26. [34]

    Continual learning in 3d point clouds: Employ- ing spectral techniques for exemplar selection,

    Resani, H., Nasihatkon, B., and Jazi, M. A., 2025, “Continual learning in 3d point clouds: Employ- ing spectral techniques for exemplar selection,” In 2025 IEEE/CVF Winter Conference on Applications of Computer Vision (W ACV), IEEE, pp. 2921–2931

    2025

  27. [35]

    On the use of surrogate models in engi- neering design optimization and exploration: The key issues,

    Palar, P. S., Liem, R. P., Zuhal, L. R., and Shimoyama, K., 2019, “On the use of surrogate models in engi- neering design optimization and exploration: The key issues,” In Proceedings of the genetic and evolutionary computation conference companion, pp. 1592–1602

    2019

  28. [36]

    Practitioners guide to mlops: A framework for con- tinuous delivery and automation of machine learning,

    Salama, K., Kazmierczak, J., and Schut, D., 2021, “Practitioners guide to mlops: A framework for con- tinuous delivery and automation of machine learning,” Google Could White paper

    2021

  29. [37]

    Ship-d: Ship hull dataset for design optimization using machine learning,

    Bagazinski, N. J., and Ahmed, F., 2023, “Ship-d: Ship hull dataset for design optimization using machine learning,” In International Design Engineering Techni- cal Conferences and Computers and Information in En- gineering Conference, V ol. 87301, American Society of Mechanica...

    2023

  30. [38]

    Bagazinski, N., and Ahmed, F., 2024, Ship-D dataset

    2024

  31. [39]

    Drivaernet++: A large-scale multimodal car dataset with computational fluid dynamics simulations and deep learning benchmarks,

    Elrefaie, M., Morar, F., Dai, A., and Ahmed, F., 2025, “Drivaernet++: A large-scale multimodal car dataset with computational fluid dynamics simulations and deep learning benchmarks,” Advances in Neural In- formation Processing Systems, 37, pp. 499–536

    2025

  32. [40]

    Jones, M., Spreizer, S., Cubra, C., and Crouse, J., 2024, Rapid aerodynamic analysis through deep learn- ing (ti95-2625 final report) Technical report, Defense Technical Information Center (DTIC) DTIC Report

    2024

  33. [41]

    Pointnet: Deep learning on point sets for 3d classi- fication and segmentation,

    Qi, C. R., Su, H., Mo, K., and Guibas, L. J., 2017, “Pointnet: Deep learning on point sets for 3d classi- fication and segmentation,” In Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 652–660

    2017

  34. [42]

    Qian, G., n.d., Pointnext model zoo Avail- able at https://guochengqian.github.io/ PointNeXt/modelzoo/

  35. [43]

    Deep residual learning for image recognition,

    He, K., Zhang, X., Ren, S., and Sun, J., 2016, “Deep residual learning for image recognition,” In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pp. 770–778

    2016

  36. [44]

    Avalanche: an end-to-end library for continual learn- ing,

    Lomonaco, V ., Pellegrini, L., Cossu, A., Carta, A., Graffieti, G., Hayes, T. L., De Lange, M., Masana, M., Pomponi, J., Van de Ven, G. M., et al., 2021, “Avalanche: an end-to-end library for continual learn- ing,” In Proceedings of the IEEE/CVF conference on computer vision a...

    2021

Pith tools

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