{"id":17684903,"url":"https://github.com/spencerwooo/torchattack","last_synced_at":"2025-04-14T22:09:29.130Z","repository":{"id":98128179,"uuid":"593526030","full_name":"spencerwooo/torchattack","owner":"spencerwooo","description":"🛡 A curated list of adversarial attacks in PyTorch, with a focus on transferable black-box attacks.","archived":false,"fork":false,"pushed_at":"2025-03-09T08:09:00.000Z","size":2033,"stargazers_count":53,"open_issues_count":0,"forks_count":5,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-03-28T10:11:29.282Z","etag":null,"topics":["adversarial-attacks","adversarial-examples","pytorch"],"latest_commit_sha":null,"homepage":"http://docs.swo.moe/torchattack/","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/spencerwooo.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null},"funding":{"custom":["https://afdian.com/a/spencerwoo","https://spencerwoo.com/sponsor"]}},"created_at":"2023-01-26T08:14:58.000Z","updated_at":"2025-03-28T09:09:43.000Z","dependencies_parsed_at":"2024-05-08T14:31:02.481Z","dependency_job_id":"d004a8e0-b25b-4d56-a8ea-a4c3e5dd0d13","html_url":"https://github.com/spencerwooo/torchattack","commit_stats":null,"previous_names":["spencerwooo/torchattack","daisylab-bit/torchattack"],"tags_count":26,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/spencerwooo%2Ftorchattack","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/spencerwooo%2Ftorchattack/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/spencerwooo%2Ftorchattack/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/spencerwooo%2Ftorchattack/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/spencerwooo","download_url":"https://codeload.github.com/spencerwooo/torchattack/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":248968914,"owners_count":21191162,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["adversarial-attacks","adversarial-examples","pytorch"],"created_at":"2024-10-24T10:25:28.397Z","updated_at":"2025-04-14T22:09:29.103Z","avatar_url":"https://github.com/spencerwooo.png","language":"Python","funding_links":["https://afdian.com/a/spencerwoo","https://spencerwoo.com/sponsor"],"categories":[],"sub_categories":[],"readme":"\u003cdiv align=\"center\"\u003e\n  \u003cp\u003e\u003cbr\u003e\u003cimg src=\"docs/images/torchattack.png\" alt=\"torchattack banner\" width=\"600\" /\u003e\u003c/p\u003e\n\u003c/div\u003e\n\n---\n\n[![Ruff](https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/astral-sh/ruff/refs/heads/main/assets/badge/v2.json)](https://github.com/astral-sh/ruff)\n[![pypi python versions](https://img.shields.io/pypi/pyversions/torchattack.svg?logo=pypi\u0026logoColor=white\u0026labelColor=2D3339)](https://pypi.python.org/pypi/torchattack)\n[![pypi version](https://img.shields.io/pypi/v/torchattack.svg?logo=pypi\u0026logoColor=white\u0026labelColor=2D3339)](https://pypi.python.org/pypi/torchattack)\n[![pypi weekly downloads](https://img.shields.io/pypi/dm/torchattack?logo=pypi\u0026logoColor=white\u0026labelColor=2D3339)](https://pypi.python.org/pypi/torchattack)\n[![lint](https://github.com/spencerwooo/torchattack/actions/workflows/ci.yml/badge.svg)](https://github.com/spencerwooo/torchattack/actions/workflows/ci.yml)\n\n🛡 **torchattack** - _A curated list of adversarial attacks in PyTorch, with a focus on transferable black-box attacks._\n\n```shell\npip install torchattack\n```\n\n## Highlights\n\n- 🛡️ A curated collection of adversarial attacks implemented in PyTorch.\n- 🔍 Focuses on gradient-based transferable black-box attacks.\n- 📦 Easily load pretrained models from torchvision or timm using `AttackModel`.\n- 🔄 Simple interface to initialize attacks with `create_attack`.\n- 🔧 Extensively typed for better code quality and safety.\n- 📊 Tooling for fooling rate metrics and model evaluation in `eval`.\n- 🔁 Numerous attacks reimplemented for readability and efficiency (TGR, VDC, etc.).\n\n## Documentation\n\ntorchattack's docs are available at [docs.swo.moe/torchattack](https://docs.swo.moe/torchattack/).\n\n## Usage\n\n```python\nimport torch\n\ndevice = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n```\n\nLoad a pretrained model to attack from either torchvision or timm.\n\n```python\nfrom torchattack import AttackModel\n\n# Load a model with `AttackModel`\nmodel = AttackModel.from_pretrained(model_name='resnet50').to(device)\n# `AttackModel` automatically attach the model's `transform` and `normalize` functions\ntransform, normalize = model.transform, model.normalize\n\n# Additionally, to explicitly specify where to load the pretrained model from (timm or torchvision),\n# prepend the model name with 'timm/' or 'tv/' respectively, or use the `from_timm` argument, e.g.\nvit_b16 = AttackModel.from_pretrained(model_name='timm/vit_base_patch16_224').to(device)\ninv_v3 = AttackModel.from_pretrained(model_name='tv/inception_v3').to(device)\npit_b = AttackModel.from_pretrained(model_name='pit_b_224', from_timm=True).to(device)\n```\n\nInitialize an attack by importing its attack class.\n\n```python\nfrom torchattack import FGSM, MIFGSM\n\n# Initialize an attack\nadversary = FGSM(model, normalize, device)\n\n# Initialize an attack with extra params\nadversary = MIFGSM(model, normalize, device, eps=0.03, steps=10, decay=1.0)\n```\n\nInitialize an attack by its name with `create_attack()`.\n\n```python\nfrom torchattack import create_attack\n\n# Initialize FGSM attack with create_attack\nadversary = create_attack('FGSM', model, normalize, device)\n\n# Initialize PGD attack with specific eps with create_attack\nadversary = create_attack('PGD', model, normalize, device, eps=0.03)\n\n# Initialize MI-FGSM attack with extra args with create_attack\nattack_args = {'steps': 10, 'decay': 1.0}\nadversary = create_attack('MIFGSM', model, normalize, device, eps=0.03, **attack_args)\n```\n\nCheck out [examples/](examples/mifgsm_transfer.py) and [`torchattack.evaluate.runner`](torchattack/evaluate/runner.py) for full examples.\n\n## Attacks\n\n\u003ctable\u003e\n  \u003cthead\u003e\n    \u003ctr\u003e\n      \u003cth\u003eName\u003c/th\u003e\n      \u003cth\u003eClass Name\u003c/th\u003e\n      \u003cth\u003ePublication\u003c/th\u003e\n      \u003cth\u003ePaper (Open Access)\u003c/th\u003e\n    \u003c/tr\u003e\n  \u003c/thead\u003e\n  \u003ctbody\u003e\n    \u003c!-- Gradient-based attacks --\u003e\n    \u003ctr\u003e\n      \u003cth colspan=\"4\"\u003eGradient-based attacks\u003c/th\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eFGSM\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eFGSM\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/ICLR-2015-62B959?labelColor=2D3339\" alt=\"ICLR 2015\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1412.6572\"\u003eExplaining and Harnessing Adversarial Examples\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003ePGD\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003ePGD\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/ICLR-2018-62B959?labelColor=2D3339\" alt=\"ICLR 2018\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1706.06083\"\u003eTowards Deep Learning Models Resistant to Adversarial Attacks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003ePGD (L2)\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003ePGDL2\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/ICLR-2018-62B959?labelColor=2D3339\" alt=\"ICLR 2018\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1706.06083\"\u003eTowards Deep Learning Models Resistant to Adversarial Attacks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eMI-FGSM\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eMIFGSM\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2018-1A407F?labelColor=2D3339\" alt=\"CVPR 2018\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1710.06081\"\u003eBoosting Adversarial Attacks with Momentum\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eDI-FGSM\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eDIFGSM\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2019-1A407F?labelColor=2D3339\" alt=\"CVPR 2019\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1803.06978\"\u003eImproving Transferability of Adversarial Examples with Input Diversity\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eTI-FGSM\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eTIFGSM\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2019-1A407F?labelColor=2D3339\" alt=\"CVPR 2019\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1904.02884\"\u003eEvading Defenses to Transferable Adversarial Examples by Translation-Invariant Attacks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eNI-FGSM\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eNIFGSM\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/ICLR-2020-62B959?labelColor=2D3339\" alt=\"ICLR 2020\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1908.06281\"\u003eNesterov Accelerated Gradient and Scale Invariance for Adversarial Attacks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eSI-NI-FGSM\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eSINIFGSM\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/ICLR-2020-62B959?labelColor=2D3339\" alt=\"ICLR 2020\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1908.06281\"\u003eNesterov Accelerated Gradient and Scale Invariance for Adversarial Attacks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eDR\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eDR\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2020-1A407F?labelColor=2D3339\" alt=\"CVPR 2020\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1911.11616\"\u003eEnhancing Cross-Task Black-Box Transferability of Adversarial Examples With Dispersion Reduction\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eVMI-FGSM\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eVMIFGSM\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2021-1A407F?labelColor=2D3339\" alt=\"CVPR 2021\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2103.15571\"\u003eEnhancing the Transferability of Adversarial Attacks through Variance Tuning\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eVNI-FGSM\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eVNIFGSM\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2021-1A407F?labelColor=2D3339\" alt=\"CVPR 2021\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2103.15571\"\u003eEnhancing the Transferability of Adversarial Attacks through Variance Tuning\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eAdmix\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eAdmix\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/ICCV-2021-5A428D?labelColor=2D3339\" alt=\"ICCV 2021\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2102.00436\"\u003eAdmix: Enhancing the Transferability of Adversarial Attacks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eFIA\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eFIA\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/ICCV-2021-5A428D?labelColor=2D3339\" alt=\"ICCV 2021\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2107.14185\"\u003eFeature Importance-aware Transferable Adversarial Attacks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003ePNA-PatchOut\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003ePNAPatchOut\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/AAAI-2022-C8172C?labelColor=2D3339\" alt=\"AAAI 2022\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2109.04176\"\u003eTowards Transferable Adversarial Attacks on Vision Transformers\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eNAA\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eNAA\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2022-1A407F?labelColor=2D3339\" alt=\"CVPR 2022\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2204.00008\"\u003eImproving Adversarial Transferability via Neuron Attribution-Based Attacks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eSSA\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eSSA\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/ECCV-2022-E16B4C?labelColor=2D3339\" alt=\"ECCV 2022\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2207.05382\"\u003eFrequency Domain Model Augmentation for Adversarial Attack\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eTGR\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eTGR\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2023-1A407F?labelColor=2D3339\" alt=\"CVPR 2023\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2303.15754\"\u003eTransferable Adversarial Attacks on Vision Transformers with Token Gradient Regularization\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eILPD\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eILPD\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/NeurIPS-2023-654287?labelColor=2D3339\" alt=\"NeurIPS 2023\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2304.13410\"\u003eImproving Adversarial Transferability via Intermediate-level Perturbation Decay\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eMIG\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eMIG\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/ICCV-2023-5A428D?labelColor=2D3339\" alt=\"ICCV 2023\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://openaccess.thecvf.com/content/ICCV2023/html/Ma_Transferable_Adversarial_Attack_for_Both_Vision_Transformers_and_Convolutional_Networks_ICCV_2023_paper.html\"\u003eTransferable Adversarial Attack for Both Vision Transformers and Convolutional Networks via Momentum Integrated Gradients\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eDeCoWA\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eDeCoWA\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/AAAI-2024-C8172C?labelColor=2D3339\" alt=\"AAAI 2024\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2402.03951\"\u003eBoosting Adversarial Transferability across Model Genus by Deformation-Constrained Warping\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eVDC\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eVDC\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/AAAI-2024-C8172C?labelColor=2D3339\" alt=\"AAAI 2024\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://ojs.aaai.org/index.php/AAAI/article/view/28541\"\u003eImproving the Adversarial Transferability of Vision Transformers with Virtual Dense Connection\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eBSR\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eBSR\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2024-1A407F?labelColor=2D3339\" alt=\"CVPR 2024\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2308.10299\"\u003eBoosting Adversarial Transferability by Block Shuffle and Rotation\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eL2T\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eL2T\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2024-1A407F?labelColor=2D3339\" alt=\"CVPR 2024\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2405.14077\"\u003eLearning to Transform Dynamically for Better Adversarial Transferability\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eATT\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eATT\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/NeurIPS-2024-654287?labelColor=2D3339\" alt=\"NeurIPS 2024\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://openreview.net/forum?id=sNz7tptCH6\"\u003eBoosting the Transferability of Adversarial Attack on Vision Transformer with Adaptive Token Tuning\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003c!-- Generative attacks --\u003e\n    \u003ctr\u003e\n      \u003cth colspan=\"4\"\u003eGenerative attacks\u003c/th\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eCDA\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eCDA\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/NeurIPS-2019-654287?labelColor=2D3339\" alt=\"NeurIPS 2019\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1905.11736\"\u003eCross-Domain Transferability of Adversarial Perturbations\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eLTP\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eLTP\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/NeurIPS-2021-654287?labelColor=2D3339\" alt=\"NeurIPS 2021\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://proceedings.neurips.cc/paper/2021/hash/7486cef2522ee03547cfb970a404a874-Abstract.html\"\u003eLearning Transferable Adversarial Perturbations\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eBIA\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eBIA\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/ICLR-2022-62B959?labelColor=2D3339\" alt=\"ICLR 2022\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2201.11528\"\u003eBeyond ImageNet Attack: Towards Crafting Adversarial Examples for Black-box Domains\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eGAMA\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eGAMA\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/NeurIPS-2022-654287?labelColor=2D3339\" alt=\"NeurIPS 2022\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2209.09502\"\u003eGAMA: Generative Adversarial Multi-Object Scene Attacks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003c!-- Others --\u003e\n    \u003ctr\u003e\n      \u003cth colspan=\"4\"\u003eOthers\u003c/th\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eDeepFool\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eDeepFool\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2016-1A407F?labelColor=2D3339\" alt=\"CVPR 2016\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/1511.04599\"\u003eDeepFool: A Simple and Accurate Method to Fool Deep Neural Networks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eGeoDA\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eGeoDA\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2020-1A407F?labelColor=2D3339\" alt=\"CVPR 2020\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2003.06468\"\u003eGeoDA: A Geometric Framework for Black-box Adversarial Attacks\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n    \u003ctr\u003e\n      \u003ctd\u003eSSP\u003c/td\u003e\n      \u003ctd\u003e\u003ccode\u003eSSP\u003c/code\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003cimg src=\"https://img.shields.io/badge/CVPR-2020-1A407F?labelColor=2D3339\" alt=\"CVPR 2020\"\u003e\u003c/td\u003e\n      \u003ctd\u003e\u003ca href=\"https://arxiv.org/abs/2006.04924\"\u003eA Self-supervised Approach for Adversarial Robustness\u003c/a\u003e\u003c/td\u003e\n    \u003c/tr\u003e\n  \u003c/tbody\u003e\n\u003c/table\u003e\n\n## Development\n\nOn how to install dependencies, run tests, and build documentation. See [Development - torchattack](https://docs.swo.moe/torchattack/development/).\n\n## License\n\n[MIT](LICENSE)\n\n## Related\n\n- [Harry24k/adversarial-attacks-pytorch](https://github.com/Harry24k/adversarial-attacks-pytorch)\n- [Trusted-AI/adversarial-robustness-toolbox](https://github.com/Trusted-AI/adversarial-robustness-toolbox)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fspencerwooo%2Ftorchattack","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fspencerwooo%2Ftorchattack","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fspencerwooo%2Ftorchattack/lists"}