Pith. sign in

REVIEW 3 major objections 7 minor 47 references

Contrastive Masked Autoencoders for Character-Level Open-Set Writer Identification

T0 review · 3 major / 7 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read A fused masked-autoencoder and contrastive-learning model can identify a writer from a single handwritten character when the writer was never seen in training, reaching 89.7% precision on CASIA-OLHWDB.

desk verdict A reasonable MAE+CL combination with a solid IAM benchmark, but the CASIA SOTA claim is unfalsifiable without baselines run under the same protocol. read the letter →

arxiv 2501.11895 v1 pith:2EC4MX3G submitted 2025-01-21 cs.CV cs.LG

classification cs.CVcs.LG
keywords contrastivemaskedautoencodersopen-setwriteridentificationcharacter-levelhandwritingonlinetrajectoriesrepresentationlearningCASIA-OLHWDBIAM-OnDBself-supervisedpretraining
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to show that one handwritten character is enough to identify its writer even when that writer's samples never appear in training. It argues this by combining masked autoencoding, which learns the sequential structure of strokes, with contrastive learning, which separates one writer's style from another's, and then testing whether two characters came from the same person. On the CASIA online handwriting database the system reaches 89.7% precision in this open-set setting, and 81.6% rank-1 accuracy on the IAM-OnDB English database, which the paper reports as state-of-the-art results. The practical stake is forensic attribution and document authentication from minimal handwriting evidence.

What carries the argument

The central object is CMAE, a Transformer-based encoder whose input is a single character trajectory normalized to (0,1], padded to 800 timepoints with pen-down, pen-up, and padding tokens, and segmented into patches of length 52. During pretraining, random patches are masked and a decoder reconstructs them with MSE loss; after the encoder, average pooling and a two-layer MLP projector feed a supervised contrastive loss, and a concatenated pair of embeddings goes to a Transformer-plus-MLP discriminator trained with cross-entropy. The masked reconstruction is tasked with capturing sequential stroke details, while the contrastive objective supplies the style-discrimination that makes unseen writers separable.

What would settle it

Run the strongest published online writer-id baseline under the paper's exact CASIA open-set protocol (20 unseen writers, 2 characters per writer, 100 random trials) and compare precision; if that baseline matches or exceeds 89.7% within the reported spread, the central claim that CMAE is state of the art for single-character open-set identification fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that open-set writer identification can be solved at the level of a single character by learning trajectory representations that are at once reconstructable and style-discriminative. The proposed CMAE model masks random patches of a character's coordinate trajectory, reconstructs them with a Transformer decoder, and in parallel uses a supervised contrastive loss to pull embeddings of the same writer together and push different writers apart; a final discriminator reads a pair of embeddings and decides whether they share a writer. The paper reports that this reaches 89.7% ± 7.4% precision under an open-set protocol on CASIA-OLHWDB (20 unseen writers, 2 characters each, averaged over 100 draws) and 81.6% ± 1.6% rank-1 accuracy on IAM-OnDB following the letter-level baseline protocol. Ablations show the contrastive branch is the dominant contributor, removing it drops CASIA precision from 89.7% to 6.5%, while removing the masked reconstruction costs only 4.3 points, and a 0.15 mask ratio beats 0.75.

Load-bearing premise

The state-of-the-art claim depends on the open-set protocol being a fair point of comparison: the paper measures 20 randomly chosen writers with 2 characters each over 100 trials and compares those numbers with earlier models that were evaluated under different protocols, without rerunning a baseline under the same protocol.

Editorial extensions

If this is right

  • A single character can serve as the query for open-set writer identification, so forensic or archival systems no longer need long text samples.
  • Writer identity is treated as a representation-learning problem rather than a closed-set classification problem, since the model is evaluated on writers absent from training.
  • The contrastive branch carries most of the discriminative power, so future methods can expect large precision losses if they replace it with generation-only objectives.
  • The same trajectory-patching and tokenization scheme transfers across scripts, as the model trains on Chinese characters and reports results on English IAM-OnDB data.
  • Mask ratio is a first-order hyperparameter: 0.15 masking outperforms 0.75 by roughly 14 precision points on CASIA.

Reading between the lines

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

  • I infer that the single-character capability could extend naturally to writer verification, where the question is only whether two fragments share a writer, and to historical scribe attribution when only one surviving glyph is available.
  • I infer that the large precision drop without contrastive learning points to a cheaper design: a contrastive-only system over patch embeddings might recover most of the discriminative signal, which would be a simple testable variant.
  • I infer that the 7.4-point standard deviation across 100 random writer draws means deployed precision will depend heavily on which writers are queried, so practical systems should report per-writer confidence intervals rather than a single number.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 7 minor

Summary. The paper proposes Contrastive Masked Autoencoders (CMAE), a representation-learning framework for character-level open-set writer identification from online handwriting trajectories. The model combines a masked autoencoder (MAE) reconstruction branch with a supervised contrastive-learning branch, followed by a transformer-based discriminator that decides whether two character trajectories come from the same writer. Experiments are reported on CASIA-OLHWDB and IAM-OnDB. The authors claim state-of-the-art performance on CASIA with 89.7% average precision under an open-set protocol, and 81.6% rank-1 accuracy on IAM-OnDB under Chen et al.'s protocol, with ablations on mask ratio, representation-learning pretraining, contrastive learning, MAE, and encoder depth.

Significance. If the central claims hold, the paper would be a useful step toward open-set writer identification from single characters, a practically relevant scenario for forensics and document authentication. The idea of combining reconstruction-based and contrastive self-supervision for online trajectory data is reasonable, and the ablations in Tables II, III, and IV provide useful evidence about which components matter (contrastive learning and pretraining are shown to be critical). However, the headline claim of state-of-the-art on CASIA is not currently falsifiable, because the evaluation protocol is bespoke and no published baseline is rerun under it; the significance of the 89.7% figure therefore cannot be assessed. The paper also has an internal inconsistency in the stated training loss, which must be resolved before the method is fully specified.

major comments (3)
  1. [Section IV.B.1, Abstract, Table II] The claim of state-of-the-art performance on CASIA (89.7% precision, Abstract and Table II) is not supported by the current evaluation. The protocol randomly selects 20 writers and 2 characters per writer and averages over 100 trials, which the authors acknowledge 'differs from previous research.' No prior method is evaluated under this same protocol, and the metric (binary same/different precision) differs from the rank-1 accuracy used in Table I. As a result, there is no reference point against which 89.7% can be judged as superior, and the claim is unfalsifiable. The authors should run at least one representative prior method (e.g., Chen et al. or a simple embedding baseline) under the identical CASIA protocol, or compare on a protocol already used in the literature.
  2. [Section III.C, Eq. (5), Section III.B.4] The training objective is not fully specified. The text states that the model employs three losses (reconstruction, supervised contrastive, and cross-entropy for the discriminator), but Eq. (5) defines the total loss as ℓ = λℓRE + (1−λ)ℓCL, omitting the cross-entropy term. Since the reported evaluation metrics are the discriminator's accuracy and precision, it is essential to state explicitly how ℓCE enters the optimization: is it added with a third weighting coefficient, optimized only for the discriminator while the encoder is trained with Eq. (5), or included in a different training stage? As written, the objective is ambiguous and the role of the discriminator in training is unclear.
  3. [Section III.B.1] The patch arithmetic appears inconsistent. The input trajectory is described as '800×2' and the patch size as 52, with 'n patches = 160.' If patching is applied along the time dimension, 800/52 is not an integer and does not equal 160; if the patch size is intended to be 5, the text should say so. This inconsistency affects the reproducibility of the architecture description and should be corrected with a precise definition of the patching operation (including whether patches are non-overlapping, and how the coordinate dimension is handled).
minor comments (7)
  1. [Section I, Contributions] The claim that the paper 'pioneers the use of single characters for writer-id tasks' should be softened, since prior work (Yang et al. 2015, Ref. [17]) already performed character-level writer identification in a closed-set setting; the novelty appears to be the open-set single-character setting, and the wording should say so precisely.
  2. [Section III.A] The notation in the task definition is inconsistent: 'τi and τj' is used in prose, while the formal statement says 'the goal is to learn a function f : (τi and τj) → σ' and later uses τ_n and superscripts. Please define the trajectory notation cleanly and use it consistently throughout.
  3. [Section III.B.2, Eq. (2)] In Eq. (2), the reconstructed output is described as 'Pm' while the caption says the decoder reconstructs 'unmasked patches'; this is contradictory. The reconstruction target should be the masked patches, so either the equation or the text should be corrected.
  4. [Section IV.B.2, Table II] The ablation 'CMAE w/o RLP' is presented as part of the mask-ratio ablation, but the difference between 'CMAE' and 'CMAE w/o RLP' is a pretraining stage, not a mask-ratio change. Consider separating the two factors or clarifying the experimental setup so that the effect of mask ratio is not confounded with pretraining.
  5. [Section IV.B.4, Fig. 4] The discussion of trajectory-length effects would be clearer if the authors reported whether the distances are averaged over many writer pairs and how the standard deviation varies; the current text does not state the number of pairs used for each length bin.
  6. [References] Some references are incomplete or have formatting artifacts (e.g., Ref. [22] and [42] contain 'V ol-3160' instead of 'Vol-3160'; Ref. [13] is missing the conference name). Please check the reference list against the publisher's styles.
  7. [Throughout] There are several typos and grammatical issues (e.g., 'taun' in Section III.B.1, 'Mask Auto-Encoders' in the Introduction, 'one a popular technique' in Section II.A). A careful proofreading pass is recommended.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: CMAE is an empirical MAE+CL system with no self-citation chain and no prediction that reduces by construction to its inputs; the CASIA SOTA claim is a missing-baseline weakness, not circular reasoning.

full rationale

This paper does not contain a derivation chain in which an output is defined in terms of the target quantity. It is an empirical machine-learning paper that combines two existing representation-learning ideas: Masked Autoencoders and supervised contrastive learning, then reports open-set writer-identification results. The reported 89.7% precision on CASIA is a model output under a bespoke evaluation protocol, not a quantity derived from the protocol itself. The absence of any baseline method run under the same CASIA protocol means the 'state-of-the-art' label is not demonstrated; however, this is an external-validity and falsifiability concern, not circularity, because the number is not forced by the definition of the metric. Likewise, the inconsistency between the text saying three losses are used and Eq. (5) listing only reconstruction and contrastive losses is an internal error that affects reproducibility, but it does not make the reported precision equivalent to an input. The IAM-OnDB comparison does follow a prior protocol (Chen et al.) and is presented against published rank-1 accuracies in Table I, giving the paper at least one benchmark-anchored evaluation. There are no self-citations among the references, and no 'uniqueness theorem' or prior author result is invoked to justify the architecture choice. Hyperparameter choices such as mask ratio and encoder depth are empirical selections, not fitted parameters re-labeled as predictions. Under the hard rule requiring a quoted reduction or self-citation chain to establish circularity, no such step can be identified; the appropriate finding is no significant circularity.

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

The central claim rests on standard deep learning components (Transformers, MAE, CL) and on domain assumptions about the encoding of handwriting trajectories. No new physical or conceptual entities are introduced. The main free parameters are hyperparameters that are tuned on validation data, and the patch_size entry shows an internal inconsistency.

free parameters (3)
  • mask_ratio = 0.15
    Chosen by hand as better than 0.75; this is the level of masking in the MAE, a core hyperparameter affecting the reported precision (89.7% vs 76.0% in Table II).
  • encoder_depth = 8
    Ablation shows depth strongly affects performance (16.8 to 89.7 precision in Table IV); 8 was selected because it achieved the highest precision.
  • patch_size = 52 (as stated), inconsistent with n_patches=160
    The paper states patch size 52 and n_patches 160 for an 800x2 input, which is arithmetically inconsistent. Patch segmentation is a free design choice influencing representation granularity.
assumptions (3)
  • domain assumption Transformer attention captures sequential dependencies in handwriting trajectories.
    The paper relies on the ability of Transformer blocks to encode stroke order and dynamics from tokenized coordinate sequences (Section III.B.1).
  • domain assumption Handwriting trajectories can be meaningfully represented by isolated coordinate pairs with special pen-up/pen-down/padding tokens.
    Preprocessing introduces three tokens to represent writing states; this assumes the chosen token values do not distort the distribution (Section IV.A.2).
  • domain assumption Contrastive learning on pairs of trajectories from the same writer yields discriminative writer styles.
    The CL module is central to the reported performance (ablation in Table III), and the paper assumes same-writer pairs are a valid positive signal.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Contrastive Masked Autoencoders for Character-Level Open-Set Writer Identification." pith.science (2026). https://pith.science/paper/2EC4MX3G

@misc{pith2026250111895,
  author       = {Pith},
  title        = {Pith review of: Contrastive Masked Autoencoders for Character-Level Open-Set Writer Identification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2EC4MX3G}},
  note         = {Machine review of arXiv:2501.11895}
}
read the original abstract

In the realm of digital forensics and document authentication, writer identification plays a crucial role in determining the authors of documents based on handwriting styles. The primary challenge in writer-id is the "open-set scenario", where the goal is accurately recognizing writers unseen during the model training. To overcome this challenge, representation learning is the key. This method can capture unique handwriting features, enabling it to recognize styles not previously encountered during training. Building on this concept, this paper introduces the Contrastive Masked Auto-Encoders (CMAE) for Character-level Open-Set Writer Identification. We merge Masked Auto-Encoders (MAE) with Contrastive Learning (CL) to simultaneously and respectively capture sequential information and distinguish diverse handwriting styles. Demonstrating its effectiveness, our model achieves state-of-the-art (SOTA) results on the CASIA online handwriting dataset, reaching an impressive precision rate of 89.7%. Our study advances universal writer-id with a sophisticated representation learning approach, contributing substantially to the ever-evolving landscape of digital handwriting analysis, and catering to the demands of an increasingly interconnected world.

Figures

Figures reproduced from arXiv: 2501.11895 by the authors.

Figure 1
Figure 1. This figure depicts the process for open-set scenarios, highlighting [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the CMAE model’s training methodology. [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Schematic of Our Proposed Model. STEP1: This framework starts by processing point sequences of single characters (including English, Chinese, and symbols). For instance, the trajectory for ”R” (depicted at the top in the figure) consists of coordinates (xi, yi) at each time ti, along with special tokens for pen down, pen up, and padding (not shown in this figure). The ”R” is then segmented into patches and subjected… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: A heatmap illustrating the differences in distance between two [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 42 canonical work pages

  1. [1]

    An introduction to biometric recognition,

    A. K. Jain, A. Ross, and S. Prabhakar, “An introduction to biometric recognition,” IEEE Trans. Circuits Syst. Video Technol., vol. 14, no. 1, pp. 4–20, 2004

  2. [2]

    An online writer identification system using regression-based feature normalization and codebook descrip- tors,

    V . Venugopal and S. Sundaram, “An online writer identification system using regression-based feature normalization and codebook descrip- tors,” Expert Syst. Appl. , vol. 72, pp. 196–206, 2017

  3. [3]

    Advances in online handwritten recognition in the last decades,

    T. Ghosh, S. Sen, S. M. Obaidullah, K. C. Santosh, K. Roy, and U. Pal, “Advances in online handwritten recognition in the last decades,” Comput. Sci. Rev., vol. 46, p. 100515, 2022

  4. [4]

    Writer recognition using off-line handwritten single block characters,

    A. L. Hagstrom, R. Stanikzai, J. Bigun, and F. Alonso-Fernandez, “Writer recognition using off-line handwritten single block characters,” in 2022 International Workshop on Biometrics and Forensics (IWBF) . IEEE, 2022, pp. 1–6

  5. [5]

    Gr-rnn: Global-context residual recurrent neural networks for writer identification,

    S. He and L. Schomaker, “Gr-rnn: Global-context residual recurrent neural networks for writer identification,” Pattern Recognit., vol. 117, p. 107975, 2021

  6. [6]

    Writer identification and writer retrieval using vision transformer for forensic documents,

    M. Koepf, F. Kleber, and R. Sablatnig, “Writer identification and writer retrieval using vision transformer for forensic documents,” inDocument Analysis Systems. Springer, 2022, pp. 352–366

  7. [7]

    Fragnet: Writer identification using deep fragment networks,

    S. He and L. Schomaker, “Fragnet: Writer identification using deep fragment networks,” IEEE Trans. Inf. Forensics Secur. , vol. 15, pp. 3013–3022, 2020

  8. [8]

    Handwriting analysis: Applications in person identification and forensic,

    F. Alaei and A. Alaei, “Handwriting analysis: Applications in person identification and forensic,” in Breakthroughs in Digital Biometrics and Forensics. Springer, 2022, pp. 147–165

Show all 47 references
  1. [9]

    Research on writer identification based on vision transformer,

    Z. Li and Q. Zhang, “Research on writer identification based on vision transformer,” J. Intell. Fuzzy Syst., vol. 45, no. 3, pp. 5169–5179, 2023

  2. [10]

    Self-writer: Clusterable embedding based self-supervised writer recognition from unlabeled data,

    Z. Mohammad, M. M. Kabir, M. M. Monowar, M. A. Hamid, and M. F. Mridha, “Self-writer: Clusterable embedding based self-supervised writer recognition from unlabeled data,” Mathematics, vol. 10, no. 24, p. 4796, 2022

  3. [11]

    Letter-level online writer identification,

    Z. Chen, H.-X. Yu, A. Wu, and W.-S. Zheng, “Letter-level online writer identification,” Int. J. Comput. Vis. , vol. 129, no. 5, pp. 1394–1409, 2021

  4. [12]

    Online handwrit- ing, signature and touch dynamics: Tasks and potentialapplications in the field of security and health,

    M. Faundez-Zanuy, J. Mekyska, and D. Impedovo, “Online handwrit- ing, signature and touch dynamics: Tasks and potentialapplications in the field of security and health,” Cogn. Comput. , vol. 13, no. 5, pp. 1406–1421, 2021

  5. [13]

    Msds: A large-scale chinese signature and token digit string dataset for handwriting verification,

    P. Zhang, J. Jiang, Y . Liu, and L. Jin, “Msds: A large-scale chinese signature and token digit string dataset for handwriting verification,” in Advances in neural information processing systems . Curran Associates, Inc., 2022, pp. 36 507–36 519

  6. [14]

    Dsdtw: Local representation learning with deep soft-dtw for dynamic signature verification,

    J. Jiang, S. Lai, L. Jin, and Y . Zhu, “Dsdtw: Local representation learning with deep soft-dtw for dynamic signature verification,” IEEE Trans. Inf. Forensics Secur., vol. 17, pp. 2198–2212, 2022

  7. [15]

    Deepwriterid: An end-to-end online text- independent writer identification system,

    W. Yang, L. Jin, and M. Liu, “Deepwriterid: An end-to-end online text- independent writer identification system,” IEEE Intell. Syst. , vol. 31, no. 2, pp. 45–53, 2016

  8. [16]

    A writer identification system for on-line whiteboard data,

    A. Schlapbach, M. Liwicki, and H. Bunke, “A writer identification system for on-line whiteboard data,” inPattern Recognit., vol. 41, no. 7, Jul 2008, pp. 2381–2397

  9. [17]

    Chinese character-level writer identifi- cation using path signature feature, dropstroke and deep cnn,

    W. Yang, L. Jin, and M. Liu, “Chinese character-level writer identifi- cation using path signature feature, dropstroke and deep cnn,” in 2015 13th International Conference on Document Analysis and Recognition (ICDAR), Aug 2015, pp. 546–550

  10. [18]

    Computational and performance aspects of pca-based face-recognition algorithms,

    H. Moon and P. J. Phillips, “Computational and performance aspects of pca-based face-recognition algorithms,” Perception, vol. 30, no. 3, pp. 303–321, 2001

  11. [19]

    Masked autoencoders are scalable vision learners,

    K. He, X. Chen, S. Xie, Y . Li, P. Dollar, and R. Girshick, “Masked autoencoders are scalable vision learners,” in 2022 IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR) , Jun 2022, pp. 15 979–15 988

  12. [20]

    Representation learning with contrastive predictive coding,

    A. v. d. Oord, Y . Li, and O. Vinyals, “Representation learning with contrastive predictive coding,” arXiv preprint arXiv:1807.03748, 2019. [Online]. Available: http://arxiv.org/abs/1807.03748

  13. [21]

    Review of bioinspired vision-tactile fusion perception (vtfp): From humans to humanoids,

    B. He, Q. Miao, Y . Zhou, Z. Wang, G. Li, and S. Xu, “Review of bioinspired vision-tactile fusion perception (vtfp): From humans to humanoids,” IEEE Trans. Med. Robot. Bionics , vol. 4, no. 4, pp. 875– 888, 2022

  14. [22]

    Character queries: A transformer-based approach to on-line handwritten character segmentation,

    M. Jungo, B. Wolf, A. Maksai, C. Musat, and A. Fischer, “Character queries: A transformer-based approach to on-line handwritten character segmentation,” in Proceedings of the 18th Italian Research Conference on Digital Libraries . CEUR-WS.org, 2023. [Online]. Available: https:...

  15. [23]

    Siman: Exploring self-supervised repre- sentation learning of scene text via similarity-aware normalization,

    C. Luo, L. Jin, and J. Chen, “Siman: Exploring self-supervised repre- sentation learning of scene text via similarity-aware normalization,” in 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Jun 2022, pp. 1029–1038

  16. [24]

    Self-supervised representation learning for online handwriting text classification,

    P. Mehralian, B. BabaAli, and A. Mohammadi, “Self-supervised representation learning for online handwriting text classification,” arXiv preprint arXiv:2310.06645 , 2023. [Online]. Available: http: //arxiv.org/abs/2310.06645

  17. [25]

    End-to-end online writer identification with recurrent neural network,

    X.-Y . Zhang, G.-S. Xie, C.-L. Liu, and Y . Bengio, “End-to-end online writer identification with recurrent neural network,” IEEE Trans. Hum.- Mach. Syst., vol. 47, no. 2, pp. 285–292, 2017

  18. [26]

    An end-to-end deep learning system for medieval writeridentification,

    M. Cilia, D. Impedovo, and G. Pirlo, “An end-to-end deep learning system for medieval writeridentification,” Pattern Recognit., vol. 101, p. 107172, 2020

  19. [27]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” in International Conference on Learning Representations , 2021. [Online]. Available: https: //openreview.net/forum?id=YicbFdNTTy

  20. [28]

    Escaping the big data paradigm with compact transformers,

    A. Hassani, S. Walton, N. Shah, A. Abuduweili, J. Li, and H. Shi, “Escaping the big data paradigm with compact transformers,” arXiv preprint arXiv:2104.05704, 2021

  21. [29]

    Dynamic programming algorithm optimization for spoken word recognition,

    H. Sakoe and S. Chiba, “Dynamic programming algorithm optimization for spoken word recognition,” IEEE Transactions on Acoustics, Speech, and Signal Processing , vol. 26, no. 1, pp. 43–49, February 1978

  22. [30]

    Handwriting analysis with focus on writer identifi- cation and writer retrieval,

    V . Christlein, “Handwriting analysis with focus on writer identifi- cation and writer retrieval,” Ph.D. dissertation, Friedrich-Alexander- Universit¨at Erlangen-N ¨urnberg, 2018

  23. [31]

    Text-independent writer identification on online arabic handwriting,

    M. Gargouri, S. Kanoun, and J.-M. Ogier, “Text-independent writer identification on online arabic handwriting,” in 2013 12th International Conference on Document Analysis and Recognition . IEEE, August 2013, pp. 428–432

  24. [32]

    Offline writer identification using convolutional neural network activation features,

    V . Christlein, D. Bernecker, A. Maier, and E. Angelopoulou, “Offline writer identification using convolutional neural network activation features,” in Pattern Recognition, Lecture Notes in Computer Science , J. Gall, P. Gehler, and B. Leibe, Eds., vol. 9358. Springer Internat...

  25. [33]

    Text-independent writer identification via cnn features and joint bayesian,

    Y . Tang and X. Wu, “Text-independent writer identification via cnn features and joint bayesian,” in 2016 15th International Conference on Frontiers in Handwriting Recognition (ICFHR) . IEEE, 2016, pp. 566–571

  26. [34]

    Learning features for writer retrieval and identification using triplet cnns,

    M. Keglevic, S. Fiel, and R. Sablatnig, “Learning features for writer retrieval and identification using triplet cnns,” in 2018 16th Interna- tional Conference on Frontiers in Handwriting Recognition (ICFHR) . IEEE, 2018, pp. 211–216

  27. [35]

    Offline writer identification using convo- lutional neural network and vlad descriptors,

    D. Liang, M. Wu, and Y . Hu, “Offline writer identification using convo- lutional neural network and vlad descriptors,” in Artificial Intelligence and Security, Lecture Notes in Computer Science , X. Sun, X. Zhang, Z. Xia, and E. Bertino, Eds., vol. 12736. Springer Internation...

  28. [36]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2016, pp. 770–778

  29. [37]

    Representation learning: A review and new perspectives,

    Y . Bengio, A. Courville, and P. Vincent, “Representation learning: A review and new perspectives,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 35, no. 8, pp. 1798–1828, 2013

  30. [38]

    Big self-supervised models are strong semi-supervised learners,

    T. Chen, S. Kornblith, K. Swersky, M. Norouzi, and G. Hinton, “Big self-supervised models are strong semi-supervised learners,” ArXiv Preprint ArXiv:2006.10029, 2020

  31. [39]

    A simple frame- work for contrastive learning of visual representations,

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple frame- work for contrastive learning of visual representations,” ArXiv Preprint ArXiv:2002.05709, 2020

  32. [40]

    Cmt-co: Contrastive learning with character movement task for handwritten text recogni- tion,

    X. Zhang, J. Wang, L. Jin, Y . Ren, and Y . Xue, “Cmt-co: Contrastive learning with character movement task for handwritten text recogni- tion,” in Proceedings of the Asian Conference on Computer Vision (ACCV), 2022, pp. 3104–3120

  33. [41]

    Sclair: Supervised contrastive learning for user and device independent airwriting recog- nition,

    A. Tripathi, A. K. Mondal, L. Kumar, and P. Ap, “Sclair: Supervised contrastive learning for user and device independent airwriting recog- nition,” IEEE Sens. Lett. , vol. 6, no. 2, pp. 1–4, 2022

  34. [42]

    Enhancement of scribal hands identification via self-supervised learning (extended abstract),

    L. Lastilla, “Enhancement of scribal hands identification via self-supervised learning (extended abstract),” in Proceedings of the 18th Italian Research Conference on Digital Libraries , ser. CEUR Workshop Proceedings, vol. 3160, 2022. [Online]. Available: https://ceur-ws.org/...

  35. [43]

    Self- supervised character-to-character distillation for text recognition,

    T. Guan, W. Shen, X. Yang, Q. Feng, Z. Jiang, and X. Yang, “Self- supervised character-to-character distillation for text recognition,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023, pp. 19 473–19 484

  36. [44]

    Improving masked autoencoders by learning where to mask,

    H. Chen, W. Zhang, Y . Wang, and X. Yang, “Improving masked autoencoders by learning where to mask,” 2023

  37. [45]

    Context autoencoder for self-supervised representation learning,

    X. Chen et al., “Context autoencoder for self-supervised representation learning,” Int. J. Comput. Vis. , 2023

  38. [46]

    Supervised contrastive learning,

    P. Khosla et al. , “Supervised contrastive learning,” ArXiv Preprint ArXiv:2004.11362, 2021. [Online]. Available: http://arxiv.org/abs/ 2004.11362

  39. [47]

    Casia online and of- fline chinese handwriting databases,

    C.-L. Liu, F. Yin, D.-H. Wang, and Q.-F. Wang, “Casia online and of- fline chinese handwriting databases,” in 2011 International Conference on Document Analysis and Recognition . IEEE, 2011, pp. 37–41

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.