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 Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Harnessing the Computation Redundancy in ViTs to Boost Adversarial Transferability
Authors: Jiani Liu, Zhiyuan Wang, Zeliang Zhang, Chao Huang, Susan Liang, Yunlong Tang, Chenliang Xu
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on the Image Net-1k dataset validate the effectiveness of our approach, showing that our methods significantly outperform existing baselines in both transferability and generality across diverse model architectures, including different variants of Vi Ts and mainstream Vision Large Language Models (VLLMs). |
| Researcher Affiliation | Academia | 1University of Rochester 2University of California, Santa Barbara |
| Pseudocode | No | The paper describes methods with mathematical formulations (e.g., equations 1, 2, 3, 4, 5, 6, 7, 8, 9, 10) and detailed textual explanations, but it does not include a distinct section or figure explicitly labeled as 'Pseudocode' or 'Algorithm'. |
| Open Source Code | Yes | Our code1 is publicly released at https://github.com/Trustworthy-AI-Group/ Transfer Attack under the name LL2S. 1The development version is available at https://github.com/Jennny L/Redun Attack-Vi T. |
| Open Datasets | Yes | Extensive experiments on the Image Net-1k dataset validate the effectiveness of our approach... In our paper, we use the Image Net as the studied dataset, which is under the BSD 3-Clause License. |
| Dataset Splits | No | We randomly sample 1,000 images from the Image Net1K dataset as our evaluation set. ... on the Image Net-1K dataset, we generate 1, 000 adversarial examples by attacking the surrogate model, and evaluate the adversarial transferability by attacking other models. |
| Hardware Specification | No | The paper states, 'Details are provided in the setup.' for experimental result reproducibility and compute resources. However, the provided text does not contain specific hardware details such as GPU/CPU models, processors, or memory used for running the experiments. |
| Software Dependencies | No | The paper states, 'Details are provided in the setup.' for experimental result reproducibility. However, the provided text does not explicitly list specific software dependencies with version numbers. |
| Experiment Setup | Yes | Following standard settings in prior work, set the maximum perturbation magnitude to ϵ = 16 255 with the momentum decay factor as 1 and use 10 attack steps for all methods. |