Learning From Biased Soft Labels
Authors: Hua Yuan, Yu Shi, Ning Xu, Xu Yang, Xin Geng, Yong Rui
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results demonstrate that our indicators can measure the effectiveness of biased soft labels generated by teachers or in these weakly-supervised learning paradigms. and Experiments Our experimentally investigate whether biased soft labels are still effective and whether the proposed indicators can measure the effectiveness of these soft labels, which consist of three parts. |
| Researcher Affiliation | Collaboration | 1 School of Computer Science and Engineering, Southeast University, Nanjing 210096, China 2 Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Applications (Southeast University), Ministry of Education, China 3 Lenovo Research, Beijing 100085, China |
| Pseudocode | No | The paper describes a heuristic method in Section 5.1 using equations, but it is not presented as structured pseudocode or a clearly labeled algorithm block. |
| Open Source Code | No | The paper does not provide any explicit statement about releasing source code or a link to a code repository. |
| Open Datasets | Yes | We consider three benchmark image datasets CIFAR-10, CIFAR-100 [23] and Tiny Image Net |
| Dataset Splits | Yes | Datasets are divided into training, validation, testing set in the ratio of 4:1:1. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., exact GPU/CPU models, memory amounts, or detailed computer specifications) used for running its experiments. It only mentions 'all student models are Wide Res Net28 2 architecture' but not the hardware it ran on. |
| Software Dependencies | No | The paper mentions using 'mini-batch SGD' but does not specify any software libraries or frameworks with version numbers (e.g., PyTorch 1.9, TensorFlow 2.x). |
| Experiment Setup | Yes | In all experiments, we use mini-batch SGD [36] with a batch size of 128 and a momentum of 0.9. Each model is trained with maximum epochs T = 200 and employs early stopping strategy with patience 20. and The punishment factor α1 ranges from 0 to 0.4 and the compensation factor α2 ranges from 0.9 to 1.3. The weight of the random labels α3 is from 1.6 to 2.3. |