Pith. sign in

REVIEW 4 major objections 4 minor 41 references

Personalization of Deep Learning

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

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

Pith's one-line read Personalized training can beat a generic model on both metrics

desk verdict Useful taxonomy and early-shaping idea, but the headline comparison is confounded because the baseline never trains on the target user's data. read the letter →

arxiv 1909.02803 v3 pith:HYCQ2STZ submitted 2019-09-06 cs.LG stat.ML

Johannes Schneider , Michail Vlachos This is my paper · ORCID
classification cs.LGstat.ML
keywords personalizationcurriculumlearningtransferdataaugmentationautoencoderdeepfairnesshandwritingrecognition
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

This paper tries to establish that a deep learning model can be steered toward a particular individual by changing when and how often that individual's data is used during training, or by enlarging her small dataset with similar examples from a shared pool. On a handwritten-digit recognition task with 100 users, the authors show that three curriculum schedules (early shaping, sample weighing, late shaping/transfer learning) and two autoencoder-based data-grouping methods raise accuracy on the individual's own test data compared with a baseline, usually at the cost of lower accuracy on the general dataset. The notable exception is that two configurations, early shaping with 400 initial individual-data iterations and transfer learning with 300 iterations, improve accuracy on both the individual's data and the general dataset at once. The paper also reports fairness as spread in accuracy across individuals and finds that better-performing models tend to have smaller spread. A sympathetic reader would take away that personalization and general performance are in tension, but not strictly so: a short, well-placed dose of individual data can help both.

What carries the argument

The carrying objects are the three shaping schedules and the autoencoder-based similarity measure. Shaping is a curriculum over data sources: batches are drawn entirely from either the individual's dataset or the global dataset, and the schedule determines when the individual's data dominates (early shaping), how often it appears (sample weighing), or whether it is used only for final fine-tuning (transfer learning). The similarity machinery is an autoencoder trained on all data; its 16-dimensional encoder output gives a latent representation, and Euclidean distance in that space ranks how close a global sample is to a user's sample (single-sample grouping) or how close one user's class-mean profile is to another user's (individual grouping). This lets a small personal dataset be augmented with look-alike data from the shared pool, which is what makes the augmentation 'personalized.'

What would settle it

Train the baseline on the global dataset plus the target user's own training data and evaluate on her held-out test set. If that baseline reaches or exceeds the accuracy of EarlyShape-ES-n-400 and TransferL-TL-n-300, then the reported advantage comes from exposure to the user's data, not from the curriculum schedule.

Watch

Extended reading notes

Core claim

The central claim is that personalization of a deep classifier can be achieved by curriculum shaping and by similarity-based data grouping, and that these are not merely trade-off maneuvers. Shaping changes the exposure schedule: early shaping trains only on the individual's data for a limited number of iterations before switching to the global set; sample weighing interleaves individual batches throughout training (one every n_I-th iteration); transfer learning fine-tunes a globally trained model on the individual's data. Data grouping replaces the individual's small dataset with an enlarged one, where added samples are either the most similar single samples in the global set or entire datasets of similar individuals, with similarity measured by Euclidean distance in a 16-dimensional autoencoder latent space. The paper's key experimental result is that EarlyShape with 400 initial iterations and TransferL with 300 iterations outperform the baseline on both the individual's held-out accuracy (0.983 vs 0.98 and 0.983 vs 0.98) and general accuracy (0.98 vs 0.979 and 0.979 vs 0.979), while sample weighing achieves the best personal accuracy (0.984) but significantly worse general accuracy (0.968).

Load-bearing premise

The comparison assumes that a baseline trained on the global dataset plus one individual's data and tested on a different individual is a fair control, so the gains of the shaping methods are attributed to the training schedule rather than to having seen a real user's data during training.

Editorial extensions

If this is right

  • Users can get a model tuned to their own handwriting with a short curriculum phase, without retraining from scratch for everyone.
  • If the two winning configurations replicate beyond this dataset, early and late shaping are both viable personalization strategies that do not force a personalization-versus-generalization sacrifice.
  • The more aggressively a model is shaped toward one individual (sample weighing), the better it does on that individual and the worse on the general set, so deployment needs a tunable exposure parameter.
  • Augmenting an individual's data with only the closest similar samples improves personal accuracy; adding too many or more diverse samples shifts performance back toward the general baseline.
  • Reporting minimum, maximum and standard deviation of accuracy across individuals gives a practical fairness check: models with higher average accuracy also showed less spread.

Reading between the lines

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

  • The paper's own evaluation section states that the baseline is tested on a different individual's data; that makes it an open question whether the shaping gains come from the schedule or from exposure to one real user's data, and a direct control would train the baseline on the target user's own training data.
  • The same autoencoder similarity could support privacy-preserving personalization: instead of sharing raw data, only latent representations of similar samples would need to be exchanged, though the paper does not explore this.
  • The class-count variation scenario was explicitly deferred, so the methods' behavior when individuals differ in how many samples they have per class remains an open testable question.
  • A practical extension would be to treat the number of initial or final individual-data iterations as a tunable knob and study how personal and general accuracy trade off as that number grows, which the paper only samples at a few points.
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. This paper discusses personalization of deep learning classifiers. It defines objectives (personalized-data accuracy, global-data accuracy, fairness) and proposes two families of methods: three curriculum/shaping schedules (early shaping, sample weighing, transfer learning) and two data-grouping augmentation strategies (individual grouping and single-sample grouping) driven by an autoencoder latent-space similarity. Experiments on the NIST Special Database 19 use a VGG-style CNN and an autoencoder, measuring accuracy on an individual's test data and on a global test set, plus fairness metrics. The central empirical claim is that both curriculum learning and personalized data augmentation improve performance on an individual's data, usually at the cost of global-data performance, with two shaping configurations improving both metrics.

Significance. If the empirical claims hold, the paper would offer a structured comparison of practical personalization techniques and a useful framing of the personalization/global-performance trade-off with fairness metrics. The data-grouping portion uses a reasonable baseline (augmenting the target user's data with random global samples) and reports consistent small gains on individual accuracy. The paper also makes an honest attempt to report variance across users via standard deviations and min/max spreads. However, the shaping evaluations rest on a confounded baseline that never sees the target user's training data, the statistical reporting is too sparse to support the stated significance claims, and all conclusions come from a single dataset and architecture without released code, weakening the generality of the conclusions.

major comments (4)
  1. [Section IV (Evaluation, shaping baseline)] The shaping baseline is not a valid control. The text states that for shaping, the baseline was trained on the union of DG and the data of one individual D_I', while evaluation was carried out on data from another individual D_I' with D_I' != D_I; in contrast, every shaping method (EarlyShape, SampleWeigh, TransferL) is trained on the target user's own training split. The reported improvements on Acc. on DI (e.g., ES-n-400 at .983 and TL-n-300 at .983 versus the baseline at .980) could therefore be caused simply by the model having seen examples of the target user's handwriting rather than by the curriculum schedule. The footnote that otherwise the baseline would be equivalent to the sample weighing method is not convincing: a baseline trained on the union DG ∪ D_I with uniform batching is a well-defined control and is not equivalent to SW-n-2, which deliberately oversamples D_I every second batch. The shaping experiments must be rerun against a control that is trained on the same data as the shaping methods, with only the ordering/weighing changed, before any claim about the benefit of shaping can be accepted.
  2. [Section IV (Results and Table I)] The statistical support for the main claims is insufficient. The paper reports p-value thresholds such as '< 0.1' and '< 0.01' without stating the test used, the number of paired observations, whether corrections for multiple comparisons were applied, or the effect size. Several of the headline comparisons (for example, .983 versus .980 on Acc. on DI for shaping, or .979 versus .975 for data grouping) are small absolute differences and are only claimed significant at the p < 0.1 level, which is above the conventional 0.05 threshold. Please provide full test details, confidence intervals, and a multiple-comparison-corrected analysis, and clarify whether these comparisons are across the 100 users or across models.
  3. [Section III (References to Algorithm 1)] The methods are repeatedly described as 'shown in Algorithm 1' (Section III, subsections A, B, C, and the data-grouping paragraph), but Algorithm 1 is not present in the manuscript. The pseudo-code is central to specifying the training schedules, sample-weighing rule, and the two grouping procedures, so its absence is a substantial reproducibility gap. Either the algorithm must be included, or the text must fully specify these procedures without relying on the missing reference.
  4. [Section IV (Evaluation setup and conclusions)] The paper draws broad conclusions about personalization from experiments on a single dataset (NIST Special Database 19), a single network architecture, and one variation type (hidden factors, not class-count variation). The abstract and conclusions state general claims such as 'shaping can outperform a baseline on both data' and 'data augmentation by choosing similar samples also increases performance of an individual.' These claims need support from at least one additional dataset or task, or the claims need to be explicitly scoped to the conditions tested. In addition, the absence of released code, random seeds, and exact hyperparameter values for the 'more than 1000 trained models' makes independent verification difficult.
minor comments (4)
  1. [Abstract and Index Terms] The abstract contains the typo 'curriculuum learning' and the index terms include 'Artificial,Feature Shaping, Intelligence,' which appears to be an artifact of misformatted keywords; please clean these up.
  2. [Section III (Data grouping) versus Section IV (Evaluation)] Section III says the similarity representation is obtained from an auto-encoder 'trained on all data,' while Section IV says the auto-encoder is trained on DG (and on DI only for sample weighing). Please clarify whether any part of the target user's data, including its test split, is used to train the auto-encoder, since this matters for the data-grouping evaluation.
  3. [Table I] The notation in Table I (ES-n-400, SW-n-12, TL-n-300, IG-n-5, SG-n-1500, etc.) is not explained before or in the table itself; please add a table note defining n for sample weighing, early shaping, transfer learning, and grouping, and stating the sizes of the augmented datasets for individual grouping.
  4. [Section IV (Fairness discussion)] The statement that 'Accuracy correlates with fairness' is made largely on the basis of the summarized rows in Table I; please either report a quantitative correlation (e.g., across the 100 users) or soften the claim, since the table only shows aggregate standard deviations and min/max values for a few configurations.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's claims are empirical and self-contained, with no derivation chain that reduces to its inputs.

full rationale

This paper is an empirical evaluation of shaping and data-grouping methods for personalizing deep learning models. It contains no mathematical derivation or formal theorem whose conclusion is equivalent to its assumptions by construction. The reported accuracies are measured on held-out test data under fixed experimental protocols. The auto-encoder used for similarity is trained on the global dataset, so the grouping results do not leak the target individual's test labels. Several self-citations appear in the related-work and references sections (e.g., [11], [20], [23], [27]), but none of these is load-bearing for the central empirical claims; they are background or prior applications. The most plausible concern raised by a skeptical reader is the baseline construction for shaping: the baseline is trained on global data plus data from one individual while being evaluated on a different individual's data, whereas the shaping methods train on the target individual's own data. However, this is an experimental-design and control concern about whether the comparison isolates the effect of curriculum scheduling; it is not a circularity in which a fitted parameter is renamed as a prediction, a result is imported from a self-citation chain, or an ansatz is smuggled in as an external theorem. Under the stated hard rules, statistical confounding and lack of a matched control do not constitute circularity. Therefore the appropriate circularity score is 0, with the control issue noted as a correctness risk rather than a circular step.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The central claims rest on standard supervised learning assumptions and a set of hand-chosen hyperparameters. No new physical or mathematical entities are introduced.

free parameters (5)
  • Sample weighing frequency n_I = 2 and 12 (two-thirds and one-half of batches from user data)
    Controls how often the individual's data is used; two values chosen by hand in Section IV.
  • Early shaping iterations n_ES = 40 and 400
    Number of initial training iterations on the individual's data; chosen by hand in Section IV.
  • Transfer learning iterations n_TL = 10 and 300
    Number of fine-tuning iterations on the individual's data; chosen by hand in Section IV.
  • Data grouping size n = 1500 and 6500 samples (or 5 and 21 individuals)
    Size of the augmented dataset for SingleSample and IndividualGroup; chosen by hand in Section IV.
  • Autoencoder latent dimension = 16
    Dimensionality of the latent space used for similarity; fixed by architecture in Figure 5.
assumptions (3)
  • domain assumption Inputs from individual datasets and the global dataset are drawn from the same space and contain the same classes.
    Stated in Section II as 'We assume that input samples X from each D_I and D_G are homogeneous, ie. drawn from the same space.'
  • domain assumption For the hidden-factors scenario, samples of a class from an individual are on average more similar to each other than samples from the global dataset.
    Stated in Section II under 'Hidden factors variation' and motivates data grouping.
  • domain assumption The global dataset provides valuable information beyond the individual's preferred outputs.
    Stated in Section I: 'We assume that the global training data provides valuable information beyond the individual's preferred outputs.'

how reviews work

0 comments
Cite this review

Pith. "Pith review of Personalization of Deep Learning." pith.science (2026). https://pith.science/paper/HYCQ2STZ

@misc{pith2026190902803,
  author       = {Pith},
  title        = {Pith review of: Personalization of Deep Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HYCQ2STZ}},
  note         = {Machine review of arXiv:1909.02803}
}
read the original abstract

We discuss training techniques, objectives and metrics toward personalization of deep learning models. In machine learning, personalization addresses the goal of a trained model to target a particular individual by optimizing one or more performance metrics, while conforming to certain constraints. To personalize, we investigate three methods of ``curriculum learning`` and two approaches for data grouping, i.e., augmenting the data of an individual by adding similar data identified with an auto-encoder. We show that both ``curriculuum learning'' and ``personalized'' data augmentation lead to improved performance on data of an individual. Mostly, this comes at the cost of reduced performance on a more general, broader dataset.

Figures

Figures reproduced from arXiv: 1909.02803 by the authors.

Figure 1
Figure 1. A box shows data of individuals varying in class counts or hidden [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The three forms of shaping used in this work [PITH_FULL_IMAGE:figures/full_fig_p001_2.png] view at source ↗
Figure 3
Figure 3. Similarity of data DI of individual I. For dataset D0, D1 is most similar and D3 the least. Type/Stride Filter Shape Type/Stride Filter Shape Conv/s1 3×3×1×32 Continued... Continued... Conv/s1 3×3×32×64 Conv/s1 1×1×128×128 MaxPool/s2 Pool 2×2 MaxPool/s2 Pool 2×2 Dropout/s1;p=.2 64×1 Conv/s1 1×1×128×256 Conv/s1 3×3×64×128 MaxPool/s2 Pool 2×2 Conv/s1 3×3×128×128 Dro./s1;p=.35 256×1 MaxPool/s2 Pool 2×2 FC/s1 256× nClas… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: VGG-style network architecture [18] Type/Stride Filter Shape Type/Stride Filter Shape Conv/s1 3×3×1×16 Continued... Continued... MaxPool/s2 Pool 2×2 Dense 512 Conv/s1 3×3×16×32 Reshape 2×2×128 MaxPool/s2 Pool 2×2 Conv/s1 3×3×128×128 Conv/s1 3×3×32×64 NN-Upsample 2 × 2 …
Figure 5
Figure 5. Figure 5: Auto-encoder architecture ie. about 30000 samples, not in any dataset DI ∈ DS, ie. |DG| = 30000. When personalizing for user I we utilized 70% of her data DI for training and the rest for testing. We used a VGG-style network [18] depicted in [PITH_FULL_IMAGE:figures/f…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 35 canonical work pages

  1. [1]

    What is personalization? perspectives on the design and implementation of personalization in information sys- tems,

    H. Fan and M. S. Poole, “What is personalization? perspectives on the design and implementation of personalization in information sys- tems,” Journal of Organizational Computing and Electronic Commerce, vol. 16, no. 3-4, pp. 179–202, 2006

    work page 2006

  2. [2]

    Explainable Recommendation: A Survey and New Perspectives

    Y . Zhang and X. Chen, “Explainable recommendation: A survey and new perspectives,” arXiv preprint arXiv:1804.11192 , 2018

    work page Pith review arXiv 2018

  3. [3]

    The technology behind personal digital assistants: An overview of the system architecture and key components,

    R. Sarikaya, “The technology behind personal digital assistants: An overview of the system architecture and key components,” IEEE Signal Processing Magazine, vol. 34, no. 1, 2017

    work page 2017

  4. [4]

    Deep learning in personalization of cardiovascular stents,

    Y . Lee, K. Veerubhotla, M. H. Jeong, and C. H. Lee, “Deep learning in personalization of cardiovascular stents,” Journal of cardiovascular pharmacology and therapeutics , 2019

    work page 2019

  5. [5]

    Machine learning approaches to personalize early prediction of asthma exacerbations,

    J. Finkelstein and I. cheol Jeong, “Machine learning approaches to personalize early prediction of asthma exacerbations,” Annals of the New York Academy of Sciences , vol. 1387, no. 1, p. 153, 2017

    work page 2017

  6. [6]

    Pgans: Personalized generative adver- sarial networks for ecg synthesis to improve patient-specific deep ecg classification,

    T. Golany and K. Radinsky, “Pgans: Personalized generative adver- sarial networks for ecg synthesis to improve patient-specific deep ecg classification,” in Proceedings of the AAAI Conference on Artificial Intelligence, 2019

    work page 2019

  7. [7]

    A day of great illumination: B.f. skinner’s discovery of shaping,

    G. B. Peterson, “A day of great illumination: B.f. skinner’s discovery of shaping,” Journal of the experimental analysis of behavior , vol. 82, no. 3, pp. 317–328, 2004

    work page 2004

  8. [8]

    Deep learning of representations for unsupervised and transfer learning,

    Y . Bengio, “Deep learning of representations for unsupervised and transfer learning,” in Proceedings of ICML Workshop on Unsupervised and Transfer Learning , 2012

    work page 2012

Show all 41 references
  1. [9]

    Privacy-preserving deep learning,

    R. Shokri and V . Shmatikov, “Privacy-preserving deep learning,” in Proc. of Conf. on computer and communications security , 2015

    2015

  2. [10]

    Deep learning with differential privacy,

    M. Abadi, A. Chu, I. Goodfellow, H. B. McMahan, I. Mironov, K. Talwar, and L. Zhang, “Deep learning with differential privacy,” in Proc. of SIGSAC Conf. on Computer and Communications Security , 2016

    2016

  3. [11]

    Personalized explanation in machine learning,

    J. Schneider and J. Handali, “Personalized explanation in machine learning,” European Conference on Information Systems (ECIS) , 2019

    2019

  4. [12]

    Fairness in machine learning: Lessons from political phi- losophy,

    R. Binns, “Fairness in machine learning: Lessons from political phi- losophy,” in Conference on Fairness, Accountability and Transparency, 2018, pp. 149–159

    2018

  5. [13]

    Counterfactual fair- ness,

    M. J. Kusner, J. Loftus, C. Russell, and R. Silva, “Counterfactual fair- ness,” in Advances in Neural Information Processing Systems (NIPS) , 2017

    2017

  6. [14]

    Fairness through awareness,

    C. Dwork, M. Hardt, T. Pitassi, O. Reingold, and R. Zemel, “Fairness through awareness,” inProceedings of the 3rd innovations in theoretical computer science conference , 2012, pp. 214–226

    2012

  7. [15]

    Fairness through computationally-bounded awareness,

    M. Kim, O. Reingold, and G. Rothblum, “Fairness through computationally-bounded awareness,” in Advances in Neural Informa- tion Processing Systems , 2018, pp. 4842–4852

    2018

  8. [16]

    Overcoming catastrophic forgetting in neural networks,

    J. Kirkpatrick, R. Pascanu, N. Rabinowitz, J. Veness, G. Desjardins, A. A. Rusu, K. Milan, J. Quan, T. Ramalho, A. Grabska-Barwinska et al., “Overcoming catastrophic forgetting in neural networks,” Proc. of the national academy of sciences , 2017

    2017

  9. [17]

    On the importance of initialization and momentum in deep learning,

    I. Sutskever, J. Martens, G. Dahl, and G. Hinton, “On the importance of initialization and momentum in deep learning,” in Int. Conf. on machine learning(ICML), 2013

    2013

  10. [18]

    Very deep convolutional networks for large-scale image recognition,

    K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” arXiv preprint arXiv:1409.1556 , 2014

    2014 arXiv

  11. [19]

    Wide & deep learning for recommender systems,

    H.-T. Cheng, L. Koc, J. Harmsen, T. Shaked, T. Chandra, H. Aradhye, G. Anderson, G. Corrado, W. Chai, M. Ispir et al. , “Wide & deep learning for recommender systems,” in Proc. of workshop on Deep Learning for Recommender Systems , 2016

    2016

  12. [20]

    Reconet: An interpretable neural architecture for recommender sys- tems,

    F. Fusco, M. Vlachos, V . Vasileiadis, K. Wardatzky, and J. Schneider, “Reconet: An interpretable neural architecture for recommender sys- tems,” in Int. Joint Conference on Artificial Intelligence (IJCAI , 2019

    2019

  13. [21]

    Web mining: Machine learning for web applications,

    H. Chen and M. Chau, “Web mining: Machine learning for web applications,” Annual review of information science and technology , vol. 38, no. 1, pp. 289–329, 2004

    2004

  14. [22]

    Power to the people: The role of humans in interactive machine learning,

    S. Amershi, M. Cakmak, W. B. Knox, and T. Kulesza, “Power to the people: The role of humans in interactive machine learning,” AI Magazine, vol. 35, no. 4, pp. 105–120, 2014

    2014

  15. [23]

    Human-to-ai coach: Improving human inputs to ai systems,

    J. Schneider, “Human-to-ai coach: Improving human inputs to ai systems,” in Symposium on Intelligent Data Analysis (IDA) , 2020

    2020

  16. [24]

    Per- sonalized multitask learning for predicting tomorrow’s mood, stress, and health,

    S. A. Taylor, N. Jaques, E. Nosakhare, A. Sano, and R. Picard, “Per- sonalized multitask learning for predicting tomorrow’s mood, stress, and health,” IEEE Transactions on Affective Computing , 2017

    2017

  17. [25]

    Limitations of mass customization,

    P. Zipkin, “Limitations of mass customization,” MIT Sloan management review, 2001

    2001

  18. [26]

    Mitchell, J

    R. Mitchell, J. Michalski, and T. Carbonell, Machine Learning - An Artificial Intelligence Approach . Springer, 2013

    2013

  19. [27]

    Principles of green data min- ing,

    J. Schneider, M. Basalla, and S. Seidel, “Principles of green data min- ing,” in Proc. of Hawaii International Conference on System Sciences (HICSS), 2019

    2019

  20. [28]

    Distributed representations of words and phrases and their composi- tionality,

    T. Mikolov, I. Sutskever, K. Chen, G. S. Corrado, and J. Dean, “Distributed representations of words and phrases and their composi- tionality,” in Advances in neural information processing systems , 2013, pp. 3111–3119

    2013

  21. [29]

    Small sample learning in big data era,

    J. Shu, Z. Xu, and D. Meng, “Small sample learning in big data era,” arXiv preprint arXiv:1808.04572 , 2018

    2018 arXiv

  22. [30]

    Transfer adaptation learning: A decade survey,

    L. Zhang, “Transfer adaptation learning: A decade survey,” arXiv preprint, 2019. [Online]. Available: https://arxiv.org/abs/1903.04687

    2019 arXiv

  23. [31]

    Curriculum learning,

    Y . Bengio, J. Louradour, R. Collobert, and J. Weston, “Curriculum learning,” in Proc. of international conference on machine learning (ICML), 2009

    2009

  24. [32]

    Deep multimodal learning: A survey on recent advances and trends,

    D. Ramachandram and G. W. Taylor, “Deep multimodal learning: A survey on recent advances and trends,” IEEE Signal Processing Magazine, vol. 34, no. 6, pp. 96–108, 2017

    2017

  25. [33]

    Label-free supervision of neural networks with physics and domain knowledge,

    R. Stewart and S. Ermon, “Label-free supervision of neural networks with physics and domain knowledge,” in Thirty-First AAAI Conference on Artificial Intelligence , 2017

    2017

  26. [34]

    Growing a brain: Fine- tuning by increasing model capacity,

    Y .-X. Wang, D. Ramanan, and M. Hebert, “Growing a brain: Fine- tuning by increasing model capacity,” in Proc. of Conference on Computer Vision and Pattern Recognition (CVPR) , 2017

    2017

  27. [35]

    Correcting sample selection bias by unlabeled data,

    J. Huang, A. Gretton, K. Borgwardt, B. Sch ¨olkopf, and A. J. Smola, “Correcting sample selection bias by unlabeled data,” in Advances in neural information processing systems , 2007

    2007

  28. [36]

    Personalization without user interruption: Boosting activity recognition in new subjects using un- labeled data,

    R. Fallahzadeh and H. Ghasemzadeh, “Personalization without user interruption: Boosting activity recognition in new subjects using un- labeled data,” in Proceedings of the 8th International Conference on Cyber-Physical Systems, 2017, pp. 293–302

    2017

  29. [37]

    Automated curriculum learning for neural networks,

    A. Graves, M. G. Bellemare, J. Menick, R. Munos, and K. Kavukcuoglu, “Automated curriculum learning for neural networks,” in Proc. of Int. Conf. on Machine Learning (ICML) , 2017

    2017

  30. [38]

    Learning by Asking Questions,

    I. Misra, R. Girshick, R. Fergus, M. Hebert, A. Gupta, and L. van der Maaten, “Learning by Asking Questions,” in Conf. on Computer Vision and Pattern Recognition (CVPR) , 2018

    2018

  31. [39]

    Self-paced learning with diversity,

    L. Jiang, D. Meng, S.-I. Yu, Z. Lan, S. Shan, and A. Hauptmann, “Self-paced learning with diversity,” inAdvances in Neural Information Processing Systems (NIPS) , 2014

    2014

  32. [40]

    Unsupervised domain adaptation for robust speech recognition via variational autoencoder-based data augmentation,

    W.-N. Hsu, Y . Zhang, and J. Glass, “Unsupervised domain adaptation for robust speech recognition via variational autoencoder-based data augmentation,” in Automatic Speech Recognition and Understanding Workshop (ASRU), 2017

    2017

  33. [41]

    Autoencoder-based semi- supervised curriculum learning for out-of-domain speaker verification,

    S. Zheng, G. Liu, H. Suo, and Y . Lei, “Autoencoder-based semi- supervised curriculum learning for out-of-domain speaker verification,” System, vol. 3, p. 98, 2019

    2019

Pith tools

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