\begin{table}[t]
\caption{Representative studies related to noisy labels or noisy supervision, 2023--2024.}\label{tab:recent23_24}
\small
\begin{tabular}{@{}L{0.23\textwidth}L{0.15\textwidth}L{0.22\textwidth}L{0.30\textwidth}@{}}
\toprule
Paper & Venue/year & Main setting & Role in this review \\
\midrule
DISC \citep{li2023disc} & CVPR 2023 & Closed-set LNL & Instance-specific memorization and dynamic selection/correction \\
Fine-grained LNL \citep{wei2023finegrained} & CVPR 2023 & Fine-grained classification & Semantic ambiguity and representation-preserving contrastive learning \\
Noisy correspondence \citep{han2023noisycorrespondence} & CVPR 2023 & Cross-modal retrieval & Noisy supervision as mismatched correspondence \\
RankMatch \citep{zhang2023rankmatch} & ICCV 2023 & Sample selection & Confidence and rank consistency beyond small-loss selection \\
Intrinsically long-tailed LNL \citep{lu2023longtailed} & ICCV 2023 & Long-tailed LNL & Coupling between imbalance, difficulty, and label reliability \\
Method-choice analysis \citep{yao2023which} & ICML 2023 & Method assumptions & Causal assumptions behind SSL-style and noise-modeling approaches \\
LogitClip \citep{wei2023logitclip} & ICML 2023 & Robust objective & Optimization-level control of memorization \\
CSOT \citep{chang2023csot} & NeurIPS 2023 & Structured selection/correction & Global and local structure-aware denoising \\
FedNoRo \citep{wu2023fednoro} & IJCAI 2023 & Federated LNL & Client heterogeneity, class imbalance, and noisy labels \\
Noisy pre-training \citep{chen2024pretraining,chen2025impact} & ICLR 2024 / TPAMI 2025 & Noisy model learning & Label noise embedded in pre-trained representations and foundation-model adaptation \\
Structural labels \citep{kim2024structural} & CVPR 2024 & Structure-aware LNL & Reverse kNN structural targets for representation geometry \\
L2B \citep{zhou2024l2b} & CVPR 2024 & Bootstrapping & Joint instance and label weighting for robust self-bootstrapping \\
VLM noisy-label detector \citep{wei2024vlmnl} & NeurIPS 2024 & VLM fine-tuning & External semantic priors for noisy-label detection \\
\botrule
\end{tabular}
\end{table}