Fine-tuning Pre-trained Models for Robustness under Noisy Labels

Authors: Sumyeong Ahn, Sihyeon Kim, Jongwoo Ko, Se-Young Yun

IJCAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We present the results of extensive testing and demonstrate both efficient and improved denoising performance on various benchmarks, surpassing previous methods. 5 Experiments In this section, we present empirical evaluations that showcase the superior performance of TURN.
Researcher Affiliation Academia Sumyeong Ahn1 , Sihyeon Kim2 , Jongwoo Ko2 and Se-Young Yun2 1Michigan State University 2Korea Advanced Institute of Science and Technology sumyeong@msu.edu, {sihk,jongwoo.ko,yunseyoung}@kaist.ac.kr,
Pseudocode Yes Algorithm 1 Pseudo code of TURN
Open Source Code No The paper does not provide an explicit statement about the release of its source code or a link to a code repository.
Open Datasets Yes Datasets. We conduct an evaluation on the synthetically noised CIFAR-100 dataset, and real-world noisy labeled dataset, the Clothing 1M and Web Vision datasets.
Dataset Splits No The paper mentions training and testing datasets, and various datasets like CIFAR-100, Clothing1M, and Web Vision, but does not provide explicit details about the specific training/validation/test splits (e.g., percentages, sample counts, or references to predefined splits) used for reproduction.
Hardware Specification No The paper discusses computational costs and training time but does not specify any particular hardware components (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions using the 'Ray [Liaw et al., 2018] hyperparameter tuning tool' but does not specify its version number or any other software dependencies with version details (e.g., Python, PyTorch/TensorFlow versions).
Experiment Setup Yes To optimize the hyperparameters for each model, we utilize the Ray [Liaw et al., 2018] hyperparameter tuning tool. This allows us to identify the appropriate settings for parameters such as learning rate, weight decay, optimizer, and batch size. For each PTM, specific optimized hyperparameters and their search spaces are described in the Appendix A. Regarding the hyperparameter for TURN, namely GMM threshold, = 0.6... For baselines, we run 5 epochs for FFT and 20 epochs for LP, while TURN is optimized 20 epochs of LP with 4 epochs for FFT in Step 2...