Pith. sign in

REVIEW 4 major objections 4 minor 37 references

Generalized Category Discovery in Event-Centric Contexts: Latent Pattern Mining with LLMs

T0 review · 4 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read PaMA claims that LLM-extracted, label-refined event patterns realign clusters with human classes in long, imbalanced event narratives, yielding up to 12.58% H-score gains over prior GCD methods

desk verdict A useful EC-GCD benchmark and a sensible LLM-pattern pipeline, but the Scam Report headline is inflated by test-set selection of rho. read the letter →

arxiv 2505.23304 v1 pith:EWJQG4JA submitted 2025-05-29 cs.CL

Yi Luo , Qiwen Wang , Junqi Yang , Luyao Tang , Zhenghao Lin , Zhenzhe Ying , Weiqiang Wang , Chen Lin This is my paper · ORCID
classification cs.CL
keywords GeneralizedCategoryDiscoveryEvent-CentricGCDLLMpatternminingclassimbalancepseudo-labelreassignmentprototypelearningscamdetectioncluster-classalignment
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 argues that Generalized Category Discovery fails on event-centric texts such as police fraud transcripts and scam complaints, because representation-based clustering groups by surface wording while humans classify by latent event structure, and because rare classes are swallowed by dominant ones. It introduces PaMA, which uses a large language model to write a textual pattern for each cluster, then refines those patterns against labeled samples so they encode human annotation criteria; the patterns reassign low-confidence samples and shape prototypes. On two event-centric datasets, including a newly built Scam Report benchmark, PaMA reports up to 12.58% higher H-score than prior GCD methods, with balanced known/novel accuracy and competitive results on standard benchmarks. If correct, this means LLM-written pattern descriptions are a workable bridge between unsupervised clusters and human category definitions in high-stakes, long-document settings.

What carries the argument

The load-bearing object is the event pattern: a short LLM-generated natural-language description of the archetypal scam or event flow behind a cluster. PaMA's ranking-filtering-mining pipeline computes a ranking score $\mathrm{Score}(C_k)=\sigma\cdot\mathrm{Comp}_k+(1-\sigma)\cdot\mathrm{Size}_k$, where compactness favors low intra-cluster variance and size favors dominant classes, so patterns for well-formed dominant clusters are generated first and minority clusters are not crowded out. Filtering removes samples already explained by existing patterns; consensus-driven extraction asks the LLM to induce a dominant pattern from remaining samples. Pattern refinement then rewrites each pattern using labeled positives and negatives, and prototypes combine statistical and semantic cues as $P_i=\beta\mu_i+(1-\beta)f_\theta(p_i)$ with EMA updates; these patterns carry the argument because they are the interface between feature-space clusters and human classification criteria.

What would settle it

Replace the refined LLM patterns with surface-keyword or random patterns while keeping the rest of PaMA fixed and re-measure H-score on Scam Report and Telecom Fraud Case: if the reported gains persist, the pattern-alignment mechanism is not the cause. A more direct check is to compare each refined pattern against held-out human annotation rationales and see whether pattern-class agreement predicts per-class accuracy gains.

Watch

Extended reading notes

Core claim

PaMA's central claim is that in event-centric GCD the cluster-class alignment problem is best solved at the level of latent event patterns rather than at the level of feature vectors. PaMA clusters unlabeled embeddings, ranks clusters by compactness and size, asks an LLM to induce a representative pattern for each cluster while filtering out samples that already match existing patterns, and then refines each pattern using labeled true positives and false positives so that patterns respect annotator boundaries. Low-confidence and unstable samples are reassigned to patterns, instance-level InfoNCE and hybrid prototype contrastive losses train the encoder, and prototypes combine class centers with pattern embeddings under exponential moving average. The paper reports H-score 50.88% on Scam Report versus 38.03% for the best baseline, 74.05% on Telecom Fraud Case, and the highest H-score on BANKING with competitive results on the other base benchmarks.

Load-bearing premise

The method assumes that the LLM can extract and refine event patterns that match human annotation criteria, and the paper does not quantitatively validate pattern quality; if the LLM's patterns diverge from annotator judgments, the pseudo-label reassignment and prototype updates would propagate those errors.

Editorial extensions

If this is right

  • If the gains hold, LLM-written pattern descriptions can serve as an interpretable, label-efficient bridge between unsupervised clusters and human category definitions in long-document domains such as fraud, legal, and clinical reporting.
  • The reported 12.58% H-score gain over the strongest baseline on Scam Report implies that previous GCD methods leave most of the difficulty in aligning clusters to subjective classes, not in representation quality alone.
  • The ablation showing that reassignment weight $\rho=25$ is optimal implies that reassigned pseudo-labels carry useful signal but must be down-weighted relative to confirmed samples to avoid injecting noise.
  • Because refinement only uses known-class labeled data, the paper's own results imply that aligning genuinely novel categories to human criteria remains unsolved and is the next bottleneck.

Reading between the lines

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

  • The pattern-as-prototype interface suggests a paper-external test: measure pattern agreement with annotator labels and use it as a predictor of per-class GCD accuracy; the paper reports no such validation.
  • Because the LLM pattern prompts are domain-agnostic, the pipeline should transfer to other event-centric corpora such as medical case notes and legal judgments, but the paper only evaluates on two fraud-domain datasets, so that transfer remains conjecture.
  • A deployment risk the paper acknowledges indirectly: detailed scam patterns readable by fraudsters could be used to evade detection, making dataset desensitization essential before release.
  • The average annotation consistency of 91.8% on Scam Report suggests label noise itself may limit the ceiling for alignment; modeling annotator disagreement explicitly could be a natural follow-up.
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 Event-Centric Generalized Category Discovery (EC-GCD), a textual GCD setting with long, complex, imbalanced, and subjectively labeled event narratives, and contributes a new Chinese Scam Report benchmark. The proposed method, PaMA, clusters unlabeled BERT embeddings, ranks clusters by compactness and size, uses Qwen2.5-72B to generate and refine event patterns, reassigns low-confidence pseudo-labels on the basis of those patterns, and trains with instance-level and prototype-level contrastive losses. Experiments compare PaMA against six GCD baselines on two EC-GCD datasets and three standard GCD benchmarks, reporting large H-score gains on EC-GCD and competitive results on base GCD.

Significance. Assuming the results survive an honest evaluation protocol, the paper makes a useful contribution: it defines a realistic EC-GCD problem, introduces a new benchmark with reported 91.8% annotator agreement, and presents a modular, clearly ablated LLM-guided alignment method. The design choices—cluster ranking, pattern refinement, pseudo-label reassignment, and the hybrid prototype objective—are each individually tested, which is a strength. The main weakness is methodological: the headline gain on Scam Report is obtained with a test-set-tuned hyperparameter, and the absence of variance estimates leaves several smaller margins indistinguishable from noise. The core idea is plausible and worth publishing after a corrected evaluation.

major comments (4)
  1. [§4.1 and §4.3.4 (Figure 4)] The Scam Report main result in Table 1 is reported with ρ=25, yet Figure 4 shows H-score peaking at ρ=25 on exactly the benchmark reported in Table 1, meaning the headline 12.58-point gain is achieved under a hyperparameter selected on the test set. Under the default ρ=1 used in the ablations, Table 2 reports H-score 43.59, which is 8.82 points above the best baseline GLEAN (34.77), not 12.58. Because the ablations use a different ρ from the main table, Table 2 is not directly comparable to Table 1. Please select hyperparameters on a held-out validation split, or make the default-ρ results primary and treat the ρ sweep as sensitivity analysis.
  2. [§4.1 vs. Table 1] §4.1 states that baselines were implemented based on their original designs with the LLM replaced for consistency, while the Table 1 caption says some results are cited from An et al. [6, 7]. These statements are mutually inconsistent. If the LOOP and GLEAN numbers are quoted from earlier papers, they may have been obtained with a different LLM, backbone, or data split, which would make the comparison to PaMA unfair. Please list exactly which cells are reproduced in this paper and which are quoted, and confirm that quoted numbers use the same evaluation protocol.
  3. [§4.1 and Table 1] The paper reports results averaged over three runs but gives no standard deviations or confidence intervals. Several claims rest on small margins, such as Telecom Fraud ACCN (+1.10) and CLINC H-score (−0.19), so without variance estimates those differences are not distinguishable from run-to-run noise. Please report per-run values or error bars for at least the main tables.
  4. [§3.2 and §4] The central mechanism—LLM-extracted patterns aligning with human annotation criteria—is never quantitatively validated. There is no measure of pattern accuracy, no human agreement study on generated patterns, and no analysis of how pattern errors propagate through pseudo-label reassignment. Such validation is needed to support the paper's attribution of performance gains to the alignment mechanism rather than to other components or to hyperparameter selection.
minor comments (4)
  1. [Introduction, Challenge ❶] The claim that the average document length in the new dataset is '14 times that of previous GCD datasets' is not substantiated anywhere; please add a table with average lengths for all datasets.
  2. [§3.2.3] The terms 'true positives and false positives' are used without defining the reference set; please specify whether these are relative to cluster assignments, current pseudo-labels, or matched known classes.
  3. [Appendix D] The subsection 'The Number of Low-Confidence Samples' refers to varying 'top-entropy samples,' while the heading and Figure 7(b) use 'low-confidence'; unifying the terminology would avoid confusion.
  4. [Figure 2] The legend entries in Figure 2 are densely packed; enlarging and separating the legend would improve readability.

Circularity Check

1 steps flagged · score 6.0 of 10

Scam Report headline gain is the maximum of a test-set rho scan, so the 12.58% margin is a fitted quantity; qualitative gains remain at default rho.

  1. fitted input called prediction [Sec. 4.1 Implementation Details and Sec. 4.3.4, Table 1 and Figure 4]
    "In our main experiments, we use the optimal sample weight of ρ = 25 for the scam report, while the ablations use the initial value of ρ = 1. ... Performance peaks at ρ = 25, with the highest H-score (50.88%) and ACCN (44.10%)."

    The reported Scam Report headline (H-score 50.88, +12.58 over GLEAN) is the value at ρ=25, and Sec. 4.3.4 identifies this ρ as the peak of an H-score scan over the same benchmark, with no validation split described. Thus the 'optimal' ρ is selected by the evaluation metric it is then used to report; the headline H-score is the maximum of the scanned values, i.e., a fitted quantity. At the default ρ=1 used for other datasets and in Table 2, the full method scores 43.59, only +8.82 over GLEAN, so the 12.58-point margin is partly an artifact of test-set selection rather than a fixed property of PaMA.

full rationale

The paper's derivation chain is otherwise empirical and self-contained: PaMA's components (LLM pattern generation, refinement with labeled data, pseudo-label reassignment, and prototype objectives) are evaluated on benchmarks without deriving the reported H-scores from the method's definitions. There is no load-bearing self-citation chain; references to prior GCD work are standard and not used to justify the central empirical claim. The principal circularity is the selection of ρ on the Scam Report evaluation set: the paper states that the main result uses the optimal ρ=25, and its own ablation shows this value maximizes H-score on the same benchmark, making the exact '12.58% gain' a fitted maximum rather than the performance of a fixed method. Because the method still outperforms baselines at the default ρ=1 (by 8.82 points on Scam Report) and the Telecom Fraud Case result uses the default ρ, the qualitative claim that PaMA improves over prior methods retains independent support. The circularity is therefore partial, affecting the precise headline margin on one dataset, which justifies a score of 6 rather than 0 or 8.

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

The central claim depends on hyperparameters tuned on the benchmark (rho=25 for Scam Report, beta, k_high, k_low from ablations) and on the untested reliability of LLM-generated patterns. No new physical or conceptual entities are introduced.

free parameters (5)
  • rho (transfer weight for reassigned pseudo-labels) = 25 for Scam Report, 1 otherwise
    Selected by sweeping rho in {0,1,5,10,20,25,30,50} and picking the value maximizing H-score on the Scam Report test set (Figure 4).
  • beta (class center ratio in prototype calculation) = 0.8
    Ablated over {0.2,0.4,0.6,0.8,1.0} and chosen for best H-score (Figure 7a).
  • k_high (number of high-confidence samples per class) = 50
    Ablated over {10,30,50,70,90} and chosen for best H-score (Figure 7c).
  • k_low (number of low-confidence samples) = 500
    Ablated over {0,250,350,450,500,650,850} and chosen for best H-score (Figure 7b).
  • sigma (compactness weight in ranking) = 0.5
    Chosen by hand as a default; no ablation reported for this value.
assumptions (3)
  • domain assumption The total number of categories K is known.
    PaMA assumes K is known (Section 3, 'we assume the total number of categories K is known'). An adaptation from Vaze et al. is mentioned but not evaluated, so this is a load-bearing assumption for the method.
  • domain assumption The LLM (Qwen2.5-72B) can reliably extract and refine event patterns that match human annotation criteria.
    Pattern generation and refinement (Section 3.2) depend on the LLM correctly identifying latent event patterns. No quantitative validation of pattern quality is provided, so this is an untested capability assumption.
  • domain assumption Cluster centers proximate in feature space correspond to identical categories (the DPN proximity assumption).
    The Hungarian matching between cluster centers and labeled class centroids (Section 3.1) relies on this assumption, following DPN [5].

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generalized Category Discovery in Event-Centric Contexts: Latent Pattern Mining with LLMs." pith.science (2026). https://pith.science/paper/EWJQG4JA

@misc{pith2026250523304,
  author       = {Pith},
  title        = {Pith review of: Generalized Category Discovery in Event-Centric Contexts: Latent Pattern Mining with LLMs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EWJQG4JA}},
  note         = {Machine review of arXiv:2505.23304}
}
read the original abstract

Generalized Category Discovery (GCD) aims to classify both known and novel categories using partially labeled data that contains only known classes. Despite achieving strong performance on existing benchmarks, current textual GCD methods lack sufficient validation in realistic settings. We introduce Event-Centric GCD (EC-GCD), characterized by long, complex narratives and highly imbalanced class distributions, posing two main challenges: (1) divergent clustering versus classification groupings caused by subjective criteria, and (2) Unfair alignment for minority classes. To tackle these, we propose PaMA, a framework leveraging LLMs to extract and refine event patterns for improved cluster-class alignment. Additionally, a ranking-filtering-mining pipeline ensures balanced representation of prototypes across imbalanced categories. Evaluations on two EC-GCD benchmarks, including a newly constructed Scam Report dataset, demonstrate that PaMA outperforms prior methods with up to 12.58% H-score gains, while maintaining strong generalization on base GCD datasets.

Figures

Figures reproduced from arXiv: 2505.23304 by the authors.

Figure 1
Figure 1. An example of question classification where the model only needs to attend to a few keywords to assign a label accurately. Challenge ❶: Divergent groupings by clus￾tering and classification. Classification relies on human-defined labels, whereas clustering is driven solely by similarity of representations. Existing textual GCD benchmarks involve struc￾turally simple short texts, where surface-level cues (e.g., salie… view at source ↗
Figure 2
Figure 2. The framework of PaMA. The pretraining stage is omitted for clarity. P(·) through the Hungarian algorithm [29]. µ l i and µ u P (i) represent matched known-class centers, while unmatched clusters are treated as novel categories. For each cluster Ck, we compute two normalized scores: Compactness, which favors low intra-cluster variance, and Size, which favors larger clusters likely representing high-frequency classes… view at source ↗
Figure 3
Figure 3. An example of Pattern Generation and Refinement. (a) Top: Sample Matching. Bottom: [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Effect of the transfer weight factor ρ. 4.3.2 Effect of Cluster Ranking and Sample Filtering To assess the role of cluster ranking and sample filtering, we ablate these components from our method. As shown in [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Results (%) of PaMA compared to baseline methods across different known class ratios. Known Class Ratio. In real-world scenarios, the proportion of known categories can vary widely. Such dynamic variations impose higher demands on the robustness and adaptability of mod…
Figure 6
Figure 6. Figure 6: The category distribution of Scam Report’s training set. [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]
Figure 7
Figure 7. Figure 7: Impact of different parameters on PaMA’s performance. (a) Ratio of the class center in prototype computation (β). (b) Number of high-confidence samples used for pattern generation. (c) Number of low-confidence samples selected for pseudo-label reassignment. The Number …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 31 canonical work pages

  1. [1]

    Generalized category discovery

    Sagar Vaze, Kai Han, Andrea Vedaldi, and Andrew Zisserman. Generalized category discovery. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022, pages 7482–7491, 2022

    work page 2022

  2. [2]

    Dynamic conceptional contrastive learning for generalized category discovery

    Nan Pu, Zhun Zhong, and Nicu Sebe. Dynamic conceptional contrastive learning for generalized category discovery. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023, Vancouver, BC, Canada, June 17-24, 2023, pages 7579–7588. IEEE, 2023

    work page 2023

  3. [3]

    Happy: A debiased learning framework for continual generalized category discovery

    Shijie Ma, Fei Zhu, Zhun Zhong, Wenzhuo Liu, Xu-Yao Zhang, and Chenglin Liu. Happy: A debiased learning framework for continual generalized category discovery. In Amir Globersons, Lester Mackey, Danielle Belgrave, Angela Fan, Ulrich Paquet, Jakub M. Tomczak, and Cheng Zhang, editors, Advances in Neural Information Processing Systems 38: Annual Conference ...

    work page 2024

  4. [4]

    Yuwei Zhang, Haode Zhang, Li-Ming Zhan, Xiao-Ming Wu, and Albert Y . S. Lam. New intent discovery with pre-training and contrastive learning. In Smaranda Muresan, Preslav Nakov, and Aline Villavicencio, editors, Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2022, Dublin, Ireland, May 2...

    work page 2022

  5. [5]

    Gen- eralized category discovery with decoupled prototypical network

    Wenbin An, Feng Tian, Qinghua Zheng, Wei Ding, Qianying Wang, and Ping Chen. Gen- eralized category discovery with decoupled prototypical network. In Brian Williams, Yiling Chen, and Jennifer Neville, editors, Thirty-Seventh AAAI Conference on Artificial Intelligence, AAAI 2023, Thirty-Fifth Conference on Innovative Applications of Artificial Intelligence...

    work page 2023

  6. [6]

    Generalized category discovery with large language models in the loop

    Wenbin An, Wenkai Shi, Feng Tian, Haonan Lin, Qianying Wang, Yaqiang Wu, Mingxiang Cai, Luyan Wang, Yan Chen, Haiping Zhu, and Ping Chen. Generalized category discovery with large language models in the loop. In Lun-Wei Ku, Andre Martins, and Vivek Srikumar, editors, Findings of the Association for Computational Linguistics, ACL 2024, Bangkok, Thailand an...

    work page 2024

  7. [7]

    Transfer and alignment network for generalized category discovery

    Wenbin An, Feng Tian, Wenkai Shi, Yan Chen, Yaqiang Wu, Qianying Wang, and Ping Chen. Transfer and alignment network for generalized category discovery. In Michael J. Wooldridge, Jennifer G. Dy, and Sriraam Natarajan, editors,Thirty-Eighth AAAI Conference on Artificial Intelligence, AAAI 2024, Thirty-Sixth Conference on Innovative Applications of Artifici...

    work page 2024

  8. [8]

    Actively learn from llms with uncertainty propagation for generalized category discovery

    Jinggui Liang, Lizi Liao, Hao Fei, Bobo Li, and Jing Jiang. Actively learn from llms with uncertainty propagation for generalized category discovery. In Kevin Duh, Helena Gómez- Adorno, and Steven Bethard, editors,Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies ...

    work page 2024

Show all 37 references
  1. [9]

    GLEAN: generalized category discovery with diverse and quality-enhanced LLM feedback

    Henry Peng Zou, Siffi Singh, Yi Nian, Jianfeng He, Jason Cai, Saab Mansour, and Hang Su. GLEAN: generalized category discovery with diverse and quality-enhanced LLM feedback. CoRR, abs/2502.18414, 2025

    2025 arXiv

  2. [10]

    Efficient intent detection with dual sentence encoders

    Iñigo Casanueva, Tadas Temcinas, Daniela Gerz, Matthew Henderson, and Ivan Vulic. Efficient intent detection with dual sentence encoders. CoRR, abs/2003.04807, 2020. 10

    2003 arXiv

  3. [11]

    Short text clustering via convolutional neural networks

    Jiaming Xu, Peng Wang, Guanhua Tian, Bo Xu, Jun Zhao, Fangyuan Wang, and Hongwei Hao. Short text clustering via convolutional neural networks. In Phil Blunsom, Shay B. Cohen, Paramveer S. Dhillon, and Percy Liang, editors, Proceedings of the 1st Workshop on Vector Space Modeli...

    2015

  4. [12]

    Peper, Christopher Clarke, Andrew Lee, Parker Hill, Jonathan K

    Stefan Larson, Anish Mahendran, Joseph J. Peper, Christopher Clarke, Andrew Lee, Parker Hill, Jonathan K. Kummerfeld, Kevin Leach, Michael A. Laurenzano, Lingjia Tang, and Jason Mars. An evaluation dataset for intent classification and out-of-scope prediction. In Kentaro Inui,...

    2019

  5. [13]

    Event-centric natural language processing

    Muhao Chen, Hongming Zhang, Qiang Ning, Manling Li, Heng Ji, Kathleen McKeown, and Dan Roth. Event-centric natural language processing. In David Chiang and Min Zhang, editors, Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th ...

    2021

  6. [14]

    Neighbor- hood contrastive learning for novel class discovery

    Zhun Zhong, Enrico Fini, Subhankar Roy, Zhiming Luo, Elisa Ricci, and Nicu Sebe. Neighbor- hood contrastive learning for novel class discovery. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10867–10875, 2021

    2021

  7. [15]

    DeCrisisMB: Debiased semi- supervised learning for crisis tweet classification via memory bank

    Henry Zou, Yue Zhou, Weizhi Zhang, and Cornelia Caragea. DeCrisisMB: Debiased semi- supervised learning for crisis tweet classification via memory bank. In Houda Bouamor, Juan Pino, and Kalika Bali, editors, Findings of the Association for Computational Linguistics: EMNLP 2023...

    2023

  8. [16]

    Watch the neighbors: A unified k-nearest neighbor contrastive learning framework for OOD intent discovery

    Yutao Mou, Keqing He, Pei Wang, Yanan Wu, Jingang Wang, Wei Wu, and Weiran Xu. Watch the neighbors: A unified k-nearest neighbor contrastive learning framework for OOD intent discovery. In Yoav Goldberg, Zornitsa Kozareva, and Yue Zhang, editors,Proceedings of the 2022 Confere...

    2022

  9. [17]

    A unified knowledge transfer network for generalized category discovery

    Wenkai Shi, Wenbin An, Feng Tian, Yan Chen, Yaqiang Wu, Qianying Wang, and Ping Chen. A unified knowledge transfer network for generalized category discovery. In Michael J. Wooldridge, Jennifer G. Dy, and Sriraam Natarajan, editors,Thirty-Eighth AAAI Conference on Artificial I...

    2024

  10. [18]

    Parametric classification for generalized category discovery: A baseline study

    Xin Wen, Bingchen Zhao, and Xiaojuan Qi. Parametric classification for generalized category discovery: A baseline study. In 2023 IEEE/CVF International Conference on Computer Vision (ICCV), pages 16544–16554, 2023

    2023

  11. [19]

    Towards distribution-agnostic generalized category discovery

    Jianhong Bai, Zuozhu Liu, Hualiang Wang, Ruizhe Chen, Lianrui Mu, Xiaomeng Li, Joey Tianyi Zhou, Y ANG FENG, Jian Wu, and Haoji Hu. Towards distribution-agnostic generalized category discovery. In A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine, editors, Ad...

    2023

  12. [20]

    Mind your outliers! investigating the negative impact of outliers on active learning for visual question answering

    Siddharth Karamcheti, Ranjay Krishna, Li Fei-Fei, and Christopher Manning. Mind your outliers! investigating the negative impact of outliers on active learning for visual question answering. In Chengqing Zong, Fei Xia, Wenjie Li, and Roberto Navigli, editors, Proceedings of th...

    2021

  13. [21]

    Mhpl: Minimum happy points learning for active source free domain adaptation

    Fan Wang, Zhongyi Han, Zhiyan Zhang, Rundong He, and Yilong Yin. Mhpl: Minimum happy points learning for active source free domain adaptation. In 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 20008–20018, 2023. 11

    2023

  14. [22]

    Deep Bayesian active learning with image data

    Yarin Gal, Riashat Islam, and Zoubin Ghahramani. Deep Bayesian active learning with image data. In Doina Precup and Yee Whye Teh, editors, Proceedings of the 34th International Conference on Machine Learning, pages 1183–1192, 2017

    2017

  15. [23]

    On the importance of effectively adapting pretrained language models for active learning

    Katerina Margatina, Loïc Barrault, and Nikolaos Aletras. On the importance of effectively adapting pretrained language models for active learning. CoRR, abs/2104.08320, 2022

    2022 arXiv

  16. [24]

    Batch active learning at scale

    Gui Citovsky, Giulia DeSalvo, Claudio Gentile, Lazaros Karydas, Anand Rajagopalan, Afshin Rostamizadeh, and Sanjiv Kumar. Batch active learning at scale. In M. Ranzato, A. Beygelzimer, Y . Dauphin, P.S. Liang, and J. Wortman Vaughan, editors,Advances in Neural Information Proc...

    2021

  17. [25]

    Active sentence learning by adversarial uncertainty sampling in discrete space

    Dongyu Ru, Jiangtao Feng, Lin Qiu, Hao Zhou, Mingxuan Wang, Weinan Zhang, Yong Yu, and Lei Li. Active sentence learning by adversarial uncertainty sampling in discrete space. In Trevor Cohn, Yulan He, and Yang Liu, editors,Findings of the Association for Computational Linguist...

    2020

  18. [26]

    Xueying Zhan, Qingzhong Wang, Kuan hao Huang, Haoyi Xiong, Dejing Dou, and Antoni B. Chan. A comparative survey of deep active learning. CoRR, abs/2203.13450, 2022

    2022 arXiv

  19. [27]

    A probabilistic framework for discovering new intents

    Yunhua Zhou, Guofeng Quan, and Xipeng Qiu. A probabilistic framework for discovering new intents. In Anna Rogers, Jordan Boyd-Graber, and Naoaki Okazaki, editors, Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), p...

    2023

  20. [28]

    ClusterPrompt: Cluster semantic enhanced prompt learning for new intent discovery

    Jinggui Liang and Lizi Liao. ClusterPrompt: Cluster semantic enhanced prompt learning for new intent discovery. In Houda Bouamor, Juan Pino, and Kalika Bali, editors, Findings of the Association for Computational Linguistics: EMNLP 2023, pages 10468–10481, 2023

    2023

  21. [29]

    Harold W. Kuhn. The hungarian method for the assignment problem. In Michael Jünger, Thomas M. Liebling, Denis Naddef, George L. Nemhauser, William R. Pulleyblank, Gerhard Reinelt, Giovanni Rinaldi, and Laurence A. Wolsey, editors,50 Years of Integer Programming 1958-2008 - Fro...

    1958

  22. [30]

    Girshick, and Ali Farhadi

    Junyuan Xie, Ross B. Girshick, and Ali Farhadi. Unsupervised deep embedding for clustering analysis. In Maria-Florina Balcan and Kilian Q. Weinberger, editors, Proceedings of the 33nd International Conference on Machine Learning, ICML 2016, New York City, NY, USA, June 19-24, ...

    2016

  23. [31]

    Representation learning with contrastive predictive coding

    Aäron van den Oord, Yazhe Li, and Oriol Vinyals. Representation learning with contrastive predictive coding. CoRR, abs/1807.03748, 2018

    2018 arXiv

  24. [32]

    Jake Snell, Kevin Swersky, and Richard S. Zemel. Prototypical networks for few-shot learning. In Isabelle Guyon, Ulrike von Luxburg, Samy Bengio, Hanna M. Wallach, Rob Fergus, S. V . N. Vishwanathan, and Roman Garnett, editors,Advances in Neural Information Processing Systems ...

    2017

  25. [33]

    Overview of CCL23-eval task 6: Telecom network fraud case classification

    Chengjie Sun, Jie Ji, Boyue Shang, and Binguan Liu. Overview of CCL23-eval task 6: Telecom network fraud case classification. In Proceedings of the 22nd Chinese National Conference on Computational Linguistics (Volume 3: Evaluations), pages 193–200, Harbin, China, August

  26. [34]

    Ovanet: One-vs-all network for universal domain adaptation

    Kuniaki Saito and Kate Saenko. Ovanet: One-vs-all network for universal domain adaptation. In 2021 IEEE/CVF International Conference on Computer Vision, ICCV 2021, Montreal, QC, Canada, October 10-17, 2021, pages 8980–8989, 2021

    2021

  27. [35]

    BERT: pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: pre-training of deep bidirectional transformers for language understanding. In Jill Burstein, Christy Doran, and Thamar Solorio, editors, Proceedings of the 2019 Conference of the North American Chapter of...

    2019

  28. [36]

    New Category

    An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, Junyang Lin, Kai Dang, Keming Lu, Keqin Bao, Kexin Yang, Le Yu, Me...

    2024 arXiv

  29. [2023]

    Chinese Information Processing Society of China

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.