Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Enhancing Robust Fairness via Confusional Spectral Regularization
Authors: Gaojie Jin, Sihao Wu, Jiaxu Liu, Tianjin Huang, Ronghui Mu
ICLR 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We validate our approach through comprehensive experiments on various datasets and models, demonstrating its effectiveness in enhancing robust fairness. To validate the effectiveness of our proposed method to improve worst-class performance, we conduct extensive experiments on various datasets. Our empirical evaluations demonstrate the superior performance of our approach in improving worst-class robust accuracy, ensuring more equitable and reliable model predictions under adversarial conditions, even for the most vulnerable classes. Empirically, extensive experiments on CIFAR10/100 and Tiny-Image Net datasets have been conducted to demonstrate the effectiveness of our method. (Sec. 4) |
| Researcher Affiliation | Academia | 1The Key Laboratory of System Software (Chinese Academy of Sciences) and State Key Laboratory of Computer Science, Institute of Software, Chinese Academy of Sciences, Beijing, China 2Department of Computer Science, University of Exeter, Exeter, UK 3Department of Computer Science, University of Liverpool, Liverpool, UK |
| Pseudocode | No | The paper describes methods through mathematical formulations and textual descriptions rather than structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | https://github.com/Alexkael/CONFUSIONAL-SPECTRAL-REGULARIZATION. |
| Open Datasets | Yes | We conduct experiments on the CIFAR-10, CIFAR-100, and Tiny Image Net datasets, which are widely used for evaluating adversarial training methods. We evaluate long-tail image classification on Image Net-LT (Liu et al., 2019) using Res Ne Xt-5032x4d as our baseline |
| Dataset Splits | Yes | We conduct experiments on the CIFAR-10, CIFAR-100, and Tiny Image Net datasets, which are widely used for evaluating adversarial training methods. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., CPU, GPU models, memory) used for the experiments. |
| Software Dependencies | No | The paper does not explicitly list specific software dependencies with version numbers (e.g., Python, PyTorch, CUDA versions). |
| Experiment Setup | Yes | We adversarially trained Preact-Res Net-18 models (He et al., 2016) for 200 epochs using SGD with a momentum of 0.9, batch size of 128, weight decay of 5 10 4, and an initial learning rate of 0.1, which is reduced by a factor of 10 at the 100th and 150th epochs. For our method, we set the value of α as 0.3, γ = 0.0/0.1, and the learning rate is 0.01. We provide sensitivity analysis for hyperparameters in App. E.1. |