Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386346346> ?p ?o ?g. }
Showing items 1 to 63 of
63
with 100 items per page.
- W4386346346 abstract "Denoising probabilistic diffusion models have shown breakthrough performance that can generate more photo-realistic images or human-level illustrations than the prior models such as GANs. This high image-generation capability has stimulated the creation of many downstream applications in various areas. However, we find that this technology is indeed a double-edged sword: We identify a new type of attack, called the Natural Denoising Diffusion (NDD) attack based on the finding that state-of-the-art deep neural network (DNN) models still hold their prediction even if we intentionally remove their robust features, which are essential to the human visual system (HVS), by text prompts. The NDD attack can generate low-cost, model-agnostic, and transferrable adversarial attacks by exploiting the natural attack capability in diffusion models. Motivated by the finding, we construct a large-scale dataset, Natural Denoising Diffusion Attack (NDDA) dataset, to systematically evaluate the risk of the natural attack capability of diffusion models with state-of-the-art text-to-image diffusion models. We evaluate the natural attack capability by answering 6 research questions. Through a user study to confirm the validity of the NDD attack, we find that the NDD attack can achieve an 88% detection rate while being stealthy to 93% of human subjects. We also find that the non-robust features embedded by diffusion models contribute to the natural attack capability. To confirm the model-agnostic and transferrable attack capability, we perform the NDD attack against an AD vehicle and find that 73% of the physically printed attacks can be detected as a stop sign. We hope that our study and dataset can help our community to be aware of the risk of diffusion models and facilitate further research toward robust DNN models." @default.
- W4386346346 created "2023-09-02" @default.
- W4386346346 creator A5002975277 @default.
- W4386346346 creator A5005366696 @default.
- W4386346346 creator A5008939086 @default.
- W4386346346 creator A5063270515 @default.
- W4386346346 creator A5066698158 @default.
- W4386346346 date "2023-08-29" @default.
- W4386346346 modified "2023-09-28" @default.
- W4386346346 title "Intriguing Properties of Diffusion Models: A Large-Scale Dataset for Evaluating Natural Attack Capability in Text-to-Image Generative Models" @default.
- W4386346346 doi "https://doi.org/10.48550/arxiv.2308.15692" @default.
- W4386346346 hasPublicationYear "2023" @default.
- W4386346346 type Work @default.
- W4386346346 citedByCount "0" @default.
- W4386346346 crossrefType "posted-content" @default.
- W4386346346 hasAuthorship W4386346346A5002975277 @default.
- W4386346346 hasAuthorship W4386346346A5005366696 @default.
- W4386346346 hasAuthorship W4386346346A5008939086 @default.
- W4386346346 hasAuthorship W4386346346A5063270515 @default.
- W4386346346 hasAuthorship W4386346346A5066698158 @default.
- W4386346346 hasBestOaLocation W43863463461 @default.
- W4386346346 hasConcept C115961682 @default.
- W4386346346 hasConcept C119857082 @default.
- W4386346346 hasConcept C121332964 @default.
- W4386346346 hasConcept C124101348 @default.
- W4386346346 hasConcept C154945302 @default.
- W4386346346 hasConcept C199360897 @default.
- W4386346346 hasConcept C2778755073 @default.
- W4386346346 hasConcept C2780801425 @default.
- W4386346346 hasConcept C37736160 @default.
- W4386346346 hasConcept C41008148 @default.
- W4386346346 hasConcept C62520636 @default.
- W4386346346 hasConcept C69357855 @default.
- W4386346346 hasConcept C97355855 @default.
- W4386346346 hasConceptScore W4386346346C115961682 @default.
- W4386346346 hasConceptScore W4386346346C119857082 @default.
- W4386346346 hasConceptScore W4386346346C121332964 @default.
- W4386346346 hasConceptScore W4386346346C124101348 @default.
- W4386346346 hasConceptScore W4386346346C154945302 @default.
- W4386346346 hasConceptScore W4386346346C199360897 @default.
- W4386346346 hasConceptScore W4386346346C2778755073 @default.
- W4386346346 hasConceptScore W4386346346C2780801425 @default.
- W4386346346 hasConceptScore W4386346346C37736160 @default.
- W4386346346 hasConceptScore W4386346346C41008148 @default.
- W4386346346 hasConceptScore W4386346346C62520636 @default.
- W4386346346 hasConceptScore W4386346346C69357855 @default.
- W4386346346 hasConceptScore W4386346346C97355855 @default.
- W4386346346 hasLocation W43863463461 @default.
- W4386346346 hasOpenAccess W4386346346 @default.
- W4386346346 hasPrimaryLocation W43863463461 @default.
- W4386346346 hasRelatedWork W2799266338 @default.
- W4386346346 hasRelatedWork W2903917280 @default.
- W4386346346 hasRelatedWork W2948798712 @default.
- W4386346346 hasRelatedWork W2961085424 @default.
- W4386346346 hasRelatedWork W2963960272 @default.
- W4386346346 hasRelatedWork W3046843850 @default.
- W4386346346 hasRelatedWork W4286629047 @default.
- W4386346346 hasRelatedWork W4300687258 @default.
- W4386346346 hasRelatedWork W4306674287 @default.
- W4386346346 hasRelatedWork W4224009465 @default.
- W4386346346 isParatext "false" @default.
- W4386346346 isRetracted "false" @default.
- W4386346346 workType "article" @default.