With False Friends Like These, Who Can Notice Mistakes?
Authors: Lue Tao, Lei Feng, Jinfeng Yi, Songcan Chen8458-8466
AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Results demonstrate the effectiveness of the countermeasures, while the risk remains non-negligible even after adaptive robust training. |
| Researcher Affiliation | Collaboration | Lue Tao1,2, Lei Feng3, Jinfeng Yi4, Songcan Chen1,2* 1College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics 2MIIT Key Laboratory of Pattern Analysis and Machine Intelligence 3College of Computer Science, Chongqing University 4JD AI Research |
| Pseudocode | No | The paper describes methods using equations and textual explanations but does not include any pseudocode or clearly labeled algorithm blocks. |
| Open Source Code | No | The paper mentions 'open-source code' in the schema context but does not explicitly state that its own methodology's code is open-source or provide a link to it. |
| Open Datasets | Yes | We evaluate the vulnerability of the substandard models across multiple network architectures... and multiple benchmark datasets, including CIFAR-10 (Krizhevsky and Hinton 2009), SVHN (Netzer et al. 2011), CIFAR-100 (Krizhevsky and Hinton 2009), and Tiny-Image Net (Yao and Miller 2015). |
| Dataset Splits | No | The paper frequently refers to 'verification data' and 'verification accuracy', implying the use of validation or test sets for evaluation. However, it does not specify the exact percentages or absolute counts for training, validation, or test splits for any of the datasets used. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., CPU, GPU models, memory, or cloud instances) used for conducting the experiments. |
| Software Dependencies | No | The paper does not list specific version numbers for any software libraries, frameworks, or programming languages used in the experiments. |
| Experiment Setup | Yes | The perturbations are generated using PGD under ℓ threat model with ϵ = 8/255 by following the common settings (Madry et al. 2018). More experimental details are provided in Appendix A. ... We set the trade-off parameter of TRADES to 6 as in (Zhang et al. 2019; Pang et al. 2021), which is too small for Tiny-Image Net. |