Pith. sign in

REVIEW 4 major objections 4 minor 27 references

Multi-Granularity Tibetan Textual Adversarial Attack Method Based on Masked Language Model

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper claims TSTricker, a masked-language-model attack on Tibetan text, cuts classifier accuracy by more than 28.70 percentage points and flips over 90.60% of test predictions.

desk verdict A solid, honest adaptation of known MLM-based attack methods to Tibetan, with a semantics-preservation claim that needs real evidence. read the letter →

arxiv 2412.02343 v1 pith:RJHCLN5E submitted 2024-12-03 cs.CL cs.CR

Xi Cao , Nuo Qun , Quzong Gesang , Yulei Zhu , Trashi Nyima This is my paper · ORCID
classification cs.CLcs.CR
keywords TibetantextualadversarialattackmaskedlanguagemodelrobustnessTibetan-BERTTiBERTtextclassificationsyllable-level
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

TSTricker is a Tibetan textual adversarial attack that generates candidate replacement syllables and words with masked language models and applies them in an order chosen by a probability-weighted saliency score. The paper claims that on Tibetan news-title classification and sentiment analysis, TSTricker lowers victim-model accuracy by more than 28.70 percentage points and changes the predicted label of more than 90.60% of test samples, outperforming the TSAttacker baseline. Syllable-level attacks, built from Tibetan-BERT's masked predictions, achieve the highest success while keeping edit distances small, and the paper reports that native speakers find their semantics unchanged. Word-level attacks, built from TiBERT's predictions, reach similar success but with larger edit distances and noticeable semantic drift. A sympathetic reader would take the paper to be establishing that publicly available masked language models suffice to expose large robustness gaps in Tibetan text classifiers.

What carries the argument

The method's load-bearing mechanism is a three-step loop run for every position in the input. First, mask a syllable or word and take the top-50 predictions of a masked language model as substitution candidates: Tibetan-BERT for syllables, TiBERT for words. Second, for each candidate compute $\Delta P = P(y|x) - P(y|x')$, the drop in the victim's probability for the original label, and keep the candidate with the largest drop. Third, order all positions by the probability-weighted word saliency $H = \mathrm{Softmax}(S) \cdot \Delta P^*$, where $S$ is the drop in classification probability when the position is replaced by an unknown token, and apply substitutions in that order until the classifier's prediction changes. The masked language models supply plausible candidates; the scoring rule decides which edits to try first.

What would settle it

A human evaluation would settle the central claim: take a sample of TSTricker-s adversarial outputs from both datasets, ask native Tibetan speakers whether each output keeps the original meaning and reads naturally, and compare the rate of 'meaning changed' or 'unnatural' judgments with the reported attack-success rates. If a large share of successful attacks are judged to have changed meaning, the attack's effectiveness comes from semantic corruption rather than imperceptible perturbation. A complementary computational check would replace the probability-weighted-saliency ordering with random ordering over the same top-50 candidate sets; if random ordering achieves similar attack-success ratios, the scoring mechanism is not the driver.

Watch

Extended reading notes

Core claim

The paper's central claim, in its own words, is that TSTricker 'can reduce the accuracy of the victim models by more than 28.70% and make the victim models change the predictions of more than 90.60% of the samples' (Section 6). Concretely, on TNCC-title the syllable-level attack lowers accuracy by 0.5124–0.5685 across four victim models and flips 99.68–99.89% of test predictions, while on TU_SA it lowers accuracy by 0.3080–0.5060 and flips 95.2–98.7% of predictions. The word-level attack is nearly as strong on accuracy drop but has much larger Levenshtein distances, and the paper concedes it changes semantics more. The paper argues that syllable-level TSTricker is the useful variant: it attacks effectively, perturbs little, preserves meaning, and can support future robustness evaluation and adversarial training.

Load-bearing premise

The load-bearing premise is that the top-50 syllables and words predicted by Tibetan-BERT and TiBERT for a masked position are natural, fluent, and preserve the sentence's meaning; if many of those candidates are unnatural or change meaning, the high attack-success numbers reflect semantic corruption rather than a clean adversarial perturbation.

Editorial extensions

If this is right

  • A robustness probe for Tibetan text classifiers can be assembled from two public masked language models, a segmentation tool, and victim-model output probabilities, with no gradient access or adversarial training required.
  • Syllable-level TSTricker supplies Tibetan adversarial examples that the paper says can seed adversarial training and robustness evaluation for Tibetan NLP models.
  • Word-level TSTricker should be reported with semantics and edit-distance metrics, since the paper finds its outputs change meaning and carry larger Levenshtein distances.
  • The same masked-candidate-plus-saliency recipe is a template for other low-resource languages that have a masked language model and a segmenter, though the paper only demonstrates it for Tibetan.

Reading between the lines

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

  • If the reported attack-success rates generalize, most of the power likely lives in the MLM candidate pool; ablating the scoring step by substituting at random positions would reveal how much the saliency ordering actually contributes.
  • Because the attack consumes the victim's class probabilities for $\Delta P$ and saliency, it is a soft-label attack; moving to a hard-label setting would require replacing the scoring step with a query-based estimator.
  • Word-level adversarial examples produced by this method may be unsafe for defense: training on them could teach a model to ignore meaningful content, whereas syllable-level examples preserve semantics and are the safer data source.
  • The paper's native-speaker judgment is reported in one sentence without counts or inter-annotator agreement; a quantified human-rating study is the natural follow-up that would test whether semantic preservation holds at scale.
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 TSTricker, a Tibetan textual adversarial attack method that operates at two granularities (syllable and word). Candidate substitutions are drawn from the top-50 masked-language-model predictions of Tibetan-BERT and TiBERT, and substitutions are ordered by a probability-weighted saliency score. The method is evaluated on two Tibetan classification datasets (TNCC-title and TU_SA) against victim models built from Tibetan-BERT and three CINO variants, with TSAttacker as the baseline. The reported results show large accuracy drops (ADV up to 0.5685) and high attack success ratios (ASR up to 0.9989), and the paper concludes that syllable-level TSTricker is a semantics-preserving, high-effectiveness attack suited for adversarial reinforcement of Tibetan NLP models.

Significance. If the results hold, this is a useful contribution to a genuinely under-studied area: adversarial robustness of Tibetan and Chinese minority language models. The paper ships public code and models, and it provides the first MLM-based multi-granularity attack for Tibetan, with a natural-language evaluation metric (LD) in addition to attack-effectiveness metrics. The quantitative attack-effectiveness results are internally consistent and substantially higher than the TSAttacker baseline. However, the central characterization of TSTricker-s as semantics-preserving is not established: the only evidence is an informal, unreported native-speaker judgment, and the paper's own example appears to contradict it. Because the paper's definition of an adversarial example requires semantics to remain basically unchanged, this missing evaluation is load-bearing for the main claim and for the recommendation to use TSTricker-s in adversarial training.

major comments (4)
  1. [Section 4.3, Table 7, Section 2.1] The claim that TSTricker-s preserves semantics is unsupported and is contradicted by the paper's own example. Section 2.1 defines an adversarial example as one whose semantics are 'basically unchanged,' and Section 6 concludes that TSTricker-s 'performs excellently in ... semantics preservation.' The only evidence is the sentence in Section 4.3 that 'native Tibetan speakers find the semantics of the TSTricker-s generated adversarial texts not changed,' with no sample size, rater agreement, or protocol. In the TNCC-title example in Table 7, the English gloss changes from 'traveling in Xizang' to 'work in Xizang' after the syllable-level substitution, which is a meaning change. A controlled human evaluation (or an automatic semantic-similarity measure with validated thresholds) is needed before the method can be called a semantics-preserving adversarial attack; otherwise the high ADV/ASR values may reflect semantic corruption rather than imperceptible perturbation.
  2. [Section 3.1, Eqs. (8)-(9)] The candidate-selection mechanism contains no semantic or fluency filter beyond the top-k MLM predictions. Equations (8)-(9) select the substitution that maximizes the drop in the victim model's probability of the true class, and the candidate pool is simply the top-50 predictions of Tibetan-BERT or TiBERT with the original token removed. This does not enforce that the chosen substitution preserves meaning, fluency, or grammaticality. Since the paper's own definition of an adversarial attack requires semantic preservation, the method as specified does not guarantee that its outputs satisfy that definition; the reported results are therefore at risk of conflating label-flipping through meaning change with true adversarial perturbation.
  3. [Tables 5-6, Section 4.3] All attack results are reported as single-run point estimates with no standard deviations, confidence intervals, or significance tests. The paper states that TSTricker has 'a significant improvement in attack effect compared with TSAttacker,' but with a single run per cell the reader cannot assess whether the differences (for example, ADV of 0.5124 vs. 0.3420 on Tibetan-BERT in Table 5) are stable or within run-to-run noise. The authors should report results over multiple fine-tuning seeds (and ideally multiple attack runs) and provide variance or significance information.
  4. [Abstract, Section 6, Tables 5-6] The abstract and conclusion state that TSTricker 'can reduce the accuracy ... by more than 28.70%' and 'change the predictions of more than 90.60% of the samples,' but the minimum ADV in Table 6 is exactly 0.2870 (TSTricker-w on TU_SA with Tibetan-BERT) and the minimum ASR is exactly 0.9060 (TSTricker-w on TU_SA with CINO-base-v2). The claims are false as stated because the table minima equal, rather than exceed, these thresholds. The wording should be corrected to 'at least' or adjusted to match the actual minima.
minor comments (4)
  1. [Section 3.2, Eq. (12)] The probability weighted word saliency of Ren et al. is applied here to syllables, but the text does not explicitly say whether the softmax normalization is taken over the syllable positions in the sentence or over the candidate set; clarifying this would make the ordering step reproducible.
  2. [Section 4.2, Eq. (15)] The Levenshtein distance definition is presented as a dynamic-programming recurrence without stating that the final LD is the value at (n, m); stating this would avoid ambiguity.
  3. [Section 4.3] The sentence describing Table 7 says that on TNCC-title both TSTricker-s and TSTricker-w change the Tibetan word for 'traveling,' but the English gloss for the TSTricker-s example shows a change to 'work'; this inconsistency should be resolved.
  4. [Section 3.1] The choice k=50 is stated as a default but no sensitivity analysis is provided; a brief discussion of how varying k affects ADV, ASR, and LD would strengthen the method.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: attack effectiveness is measured directly, not derived from fitted inputs.

full rationale

The paper's derivation chain is self-contained in the relevant sense. Candidate substitutions are generated by frozen masked language models (Tibetan-BERT and TiBERT) via top-50 predictions; substitution selection (Eqs. 8-9) maximizes the victim model's probability drop, and substitution order uses saliency (Eqs. 10-12). The reported ADV and ASR values are measured on held-out test sets of the fine-tuned victim models, not obtained by fitting a parameter to those test labels and then renaming the fit a prediction. The baseline TSAttacker and some resources (Tibetan-BERT, TNCC-title, TU_SA) originate from overlapping author groups, but the paper does not rely on accepting those prior results as premises for its attack; the comparison is to an implemented method and the central effectiveness numbers stand independently. The semantics-preservation claim in Section 4.3 rests on unreported native-speaker judgments and should be scrutinized as an empirical-support weakness, especially since the Table 7 example appears to change 'traveling' to 'work,' but this is not circularity: the claim is not defined in terms of the attack's own outputs. No equation in the paper reduces to its own inputs.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

TSTricker combines existing building blocks: masked LM candidate generation (Tibetan-BERT, TiBERT), word saliency (Li et al.), and probability-weighted word saliency (Ren et al.). The only hand-set attack hyperparameter is top-k=50. The ledger is small because the paper is an empirical attack evaluation, not a theoretical derivation; no new entities are introduced. The main unverified assumptions are that MLM candidates are semantic-preserving and that the saliency ordering helps, plus the representativeness of single-run fine-tuned victim models.

free parameters (1)
  • top-k candidate pool size k = 50 (default)
    Section 3.1 sets k=50 by default; no sensitivity analysis is reported, and the candidate pool size directly shapes the attack success rate.
assumptions (4)
  • domain assumption Top-50 masked predictions of Tibetan-BERT and TiBERT form a pool of valid, fluent, semantics-preserving substitution candidates.
    Section 3.1 builds C_s and C_w from these predictions; no quantitative fluency or semantic-similarity check is reported, only anecdotal native-speaker judgment.
  • domain assumption Probability-weighted word saliency, H = Softmax(S) * Delta-P*, is an effective ordering for Tibetan attacks.
    Section 3.2 adopts the PWWS metric from Ren et al. [16]; the paper provides no ablation showing it outperforms random or saliency-only ordering for Tibetan.
  • domain assumption Softmax probabilities P(y|x) of the fine-tuned victim models are reliable enough for candidate scoring and attack-success detection.
    Equations 6 to 12 and Algorithm 1 use P(y|x) to rank substitutions; no calibration analysis is reported.
  • domain assumption The 8:1:1 single-split fine-tuning produces representative victim models whose test-set behavior reflects real-world vulnerability.
    Section 4.1 uses one split and one seed per setup; no repeated runs or confidence intervals are provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-Granularity Tibetan Textual Adversarial Attack Method Based on Masked Language Model." pith.science (2026). https://pith.science/paper/RJHCLN5E

@misc{pith2026241202343,
  author       = {Pith},
  title        = {Pith review of: Multi-Granularity Tibetan Textual Adversarial Attack Method Based on Masked Language Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RJHCLN5E}},
  note         = {Machine review of arXiv:2412.02343}
}
read the original abstract

In social media, neural network models have been applied to hate speech detection, sentiment analysis, etc., but neural network models are susceptible to adversarial attacks. For instance, in a text classification task, the attacker elaborately introduces perturbations to the original texts that hardly alter the original semantics in order to trick the model into making different predictions. By studying textual adversarial attack methods, the robustness of language models can be evaluated and then improved. Currently, most of the research in this field focuses on English, and there is also a certain amount of research on Chinese. However, there is little research targeting Chinese minority languages. With the rapid development of artificial intelligence technology and the emergence of Chinese minority language models, textual adversarial attacks become a new challenge for the information processing of Chinese minority languages. In response to this situation, we propose a multi-granularity Tibetan textual adversarial attack method based on masked language models called TSTricker. We utilize the masked language models to generate candidate substitution syllables or words, adopt the scoring mechanism to determine the substitution order, and then conduct the attack method on several fine-tuned victim models. The experimental results show that TSTricker reduces the accuracy of the classification models by more than 28.70% and makes the classification models change the predictions of more than 90.60% of the samples, which has an evidently higher attack effect than the baseline method.

Figures

Figures reproduced from arXiv: 2412.02343 by the authors.

Figure 1
Figure 1. Structure of Tibetan Word Programming [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Process of Generating Substitutions s1 s2 s3 s4 s5 s6 s7 s8 s9 w1 w2 w3 w4 w5 w6 s* s* 1 2 s* s* 3 4 s* s* 5 6 s* s* 7 8 s*9 s*1 s2 s3 s4 s*5 s6 s7 s8 s9 w* w* 1 2 w* w* 3 4 w* w* 5 6 w1 w2 w w* 3 4 w5 w6 original text x or adversarial text x’ f(x)≠f(x’) H H ꍌ ꍍ ꍌ [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Process of Determining Substitution Order [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Performance of Victim Language Models [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: clearly depicts the attack performance [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 13 canonical work pages

  1. [1]

    Xi Cao, Dolma Dawa, Nuo Qun, and Trashi Nyima. 2023. Pay Attention to the Robustness of Chinese Minority Language Models! Syllable-level Textual Adversarial Attack on Tibetan Script. In Proceedings of the 3rd Workshop on Trustworthy Natural Language Processing (TrustNLP 2023), Anaelia Ovalle, Kai-Wei Chang, Ninareh Mehrabi, Yada Pruksachatkun, Aram Galyst...

    2023

  2. [2]

    Junjie Deng, Hanru Shi, Xinhe Yu, Wugedele Bao, Yuan Sun, and Xiaobing Zhao. 2023. MiLMo:Minority Multilingual Pre-Trained Language Model. In 2023 IEEE International Conference on Systems, Man, and Cybernetics (SMC) . 329–334. https://doi.org/10.1109/SMC53992.2023.10393961

    arXiv 2023

  3. [3]

    Javid Ebrahimi, Anyi Rao, Daniel Lowd, and Dejing Dou. 2018. HotFlip: White- Box Adversarial Examples for Text Classification. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), Iryna Gurevych and Yusuke Miyao (Eds.). Association for Computational Linguistics, Melbourne, Australia, 31–36. h...

    doi:10.18653/v1/p18-2006 2018

  4. [4]

    Steffen Eger, Gözde Gül Şahin, Andreas Rücklé, Ji-Ung Lee, Claudia Schulz, Mohsen Mesgar, Krishnkant Swarnkar, Edwin Simpson, and Iryna Gurevych

    work page

  5. [5]

    Goodfellow, Jonathon Shlens, and Christian Szegedy

    Ian J. Goodfellow, Jonathon Shlens, and Christian Szegedy. 2015. Explaining and Harnessing Adversarial Examples. In 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings, Yoshua Bengio and Yann LeCun (Eds.). http://arxiv.org/abs/1412.6572

    arXiv 2015

  6. [6]

    Ziyi Han, Wei Wang, and Shichang Xuan. 2023. Chinese Adversarial Exam- ple Generation Guided by Multi-constraints. Journal of Chinese Information Processing 37, 2 (2023), 41–52. http://jcip.cipsc.org.cn/CN/Y2023/V37/I2/41

    work page 2023

  7. [7]

    Robin Jia and Percy Liang. 2017. Adversarial Examples for Evaluating Reading Comprehension Systems. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing , Martha Palmer, Rebecca Hwa, and Sebastian Riedel (Eds.). Association for Computational Linguistics, Copenhagen, Denmark, 2021–2031. https://doi.org/10.18653/v1/D17-1215

    doi:10.18653/v1/d17-1215 2017

  8. [8]

    Di Jin, Zhijing Jin, Joey Tianyi Zhou, and Peter Szolovits. 2020. Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment. In The Thirty-Fourth AAAI Conference on Artificial Intelligence, AAAI 2020, The Thirty-Second Innovative Applications of Artificial Intelligence Conference, IAAI 2020, The Tenth AAAI ...

    2020

Show all 27 references
  1. [9]

    Bin Li, Yixuan Weng, Bin Sun, and Shutao Li. 2022. A Multi-tasking and Multi- stage Chinese Minority Pre-trained Language Model. In Machine Translation - 18th China Conference, CCMT 2022, Lhasa, China, August 6-10, 2022, Revised Selected Papers (Communications in Computer and ...

    2022 doi

  2. [10]

    Jiwei Li, Will Monroe, and Dan Jurafsky. 2016. Understanding Neural Networks through Representation Erasure. CoRR abs/1612.08220 (2016). arXiv:1612.08220 http://arxiv.org/abs/1612.08220

    2016 arXiv

  3. [11]

    Xiangge Li, Hong Luo, and Yan Sun. 2023. Adversarial Sample Generation Method Based on Chinese Features. Journal of Software 34, 11 (2023), 5143–5161. https://doi.org/10.13328/j.cnki.jos.006744

    2023 doi

  4. [12]

    Han Liu, Zhi Xu, Xiaotong Zhang, Xiaoming Xu, Feng Zhang, Fenglong Ma, Hongyang Chen, Hong Yu, and Xianchao Zhang. 2023. SSPAttack: A Simple and Sweet Paradigm for Black-Box Hard-Label Textual Adversarial Attack. In Thirty-Seventh AAAI Conference on Artificial Intelligence, AA...

    2023

  5. [13]

    Sisi Liu, Junjie Deng, Yuan Sun, and Xiaobing Zhao. 2022. TiBERT: Tibetan Pre-trained Language Model. In 2022 IEEE International Conference on Systems, Man, and Cybernetics (SMC) . 2956–2961. https://doi.org/10.1109/SMC53654.2022. 9945074

    2022

  6. [14]

    Rishabh Maheshwary, Saket Maheshwary, and Vikram Pudi. 2021. Generating Natural Language Attacks in a Hard Label Black Box Setting. InThirty-Fifth AAAI Conference on Artificial Intelligence, AAAI 2021, Thirty-Third Conference on Inno- vative Applications of Artificial Intellig...

    2021 doi

  7. [15]

    Nuo Qun, Xing Li, Xipeng Qiu, and Xuanjing Huang. 2017. End-to-End Neural Text Classification for Tibetan. In Chinese Computational Linguistics and Natural Language Processing Based on Naturally Annotated Big Data - 16th China National Conference, CCL 2017, - and - 5th Interna...

    2017

  8. [16]

    Shuhuai Ren, Yihe Deng, Kun He, and Wanxiang Che. 2019. Generating Natural Language Adversarial Examples through Probability Weighted Word Saliency. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Anna Korhonen, David Traum, and Lluí...

    2019

  9. [17]

    Goodfellow, and Rob Fergus

    Christian Szegedy, Wojciech Zaremba, Ilya Sutskever, Joan Bruna, Dumitru Erhan, Ian J. Goodfellow, and Rob Fergus. 2014. Intriguing properties of neural networks. In 2nd International Conference on Learning Representations, ICLR 2014, Banff, AB, Canada, April 14-16, 2014, Conf...

    2014 arXiv

  10. [18]

    Jindong Wang, Xixu Hu, Wenxin Hou, Hao Chen, Runkai Zheng, Yidong Wang, Linyi Yang, Haojun Huang, Wei Ye, Xiubo Geng, Binxing Jiao, Yue Zhang, and Xing Xie. 2023. On the Robustness of ChatGPT: An Adversarial and Out-of- distribution Perspective. CoRR abs/2302.12095 (2023). htt...

    2023 doi arXiv

  11. [19]

    Wenqi Wang, Run Wang, Lina Wang, and Benxiao Tang. 2019. Adversarial exam- ples generation approach for tendency classification on Chinese texts. Journal of Software 30, 8 (2019), 2415–2427. https://doi.org/10.13328/j.cnki.jos.005765

    2019 doi

  12. [20]

    Ziqing Yang, Zihang Xu, Yiming Cui, Baoxin Wang, Min Lin, Dayong Wu, and Zhigang Chen. 2022. CINO: A Chinese Minority Pre-trained Language Model. In Proceedings of the 29th International Conference on Computational Linguistics , Nicoletta Calzolari, Chu-Ren Huang, Hansaem Kim,...

    2022

  13. [21]

    Muchao Ye, Chenglin Miao, Ting Wang, and Fenglong Ma. 2022. TextHoaxer: Budgeted Hard-Label Adversarial Attacks on Text. In Thirty-Sixth AAAI Confer- ence on Artificial Intelligence, AAAI 2022, Thirty-Fourth Conference on Innovative Applications of Artificial Intelligence, IAA...

    2022 doi

  14. [22]

    Jiangyan Zhang, Deji Kazhuo, Luosang Gadeng, Nyima Trashi, and Nuo Qun

  15. [23]

    Shunxiang Zhang, Houyue Wu, Guangli Zhu, Xin Xu, and Mingxing Su. 2023. Character-level Adversarial Samples Generation Approach for Chinese Text Classification. Journal of Electronics & Information Technology 45, 6 (2023), 2226–

    2023

  16. [24]

    Yulei Zhu, Kazhuo Deji, Nuo Qun, and Tashi Nyima. 2023. Sentiment Analysis of Tibetan Short Texts Based on Graphical Neural Networks and Pre-training Models. Journal of Chinese Information Processing 37, 2 (2023), 71–79. http: //jcip.cipsc.org.cn/CN/Y2023/V37/I2/71

    2023

  17. [2019]

    Text Processing Like Humans Do: Visually Attacking and Shielding NLP Systems. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) , Jill Burstein, Ch...

    2019 doi

  18. [2022]

    In Proceedings of the 2nd International Conference on Control and Intelligent Robotics, ICCIR 2022, Nanjing, China, June 24-26, 2022

    Research and Application of Tibetan Pre-training Language Model Based on BERT. In Proceedings of the 2nd International Conference on Control and Intelligent Robotics, ICCIR 2022, Nanjing, China, June 24-26, 2022 . ACM, 519–524. https: //doi.org/10.1145/3548608.3559255

    2022

  19. [2235]

    https://doi.org/10.11999/JEIT220563

    doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.