Concept-based Adversarial Attack: a Probabilistic Perspective Article Swipe

PDF

Andi Zhang , X. X. Ding , Steven McDonagh , Samuel Kaski ·

YOU? · · 2025 · Open Access ·

We propose a concept-based adversarial attack framework that extends beyond single-image perturbations by adopting a probabilistic perspective. Rather than modifying a single image, our method operates on an entire concept -- represented by a probabilistic generative model or a set of images -- to generate diverse adversarial examples. Preserving the concept is essential, as it ensures that the resulting adversarial images remain identifiable as instances of the original underlying category or identity. By sampling from this concept-based adversarial distribution, we generate images that maintain the original concept but vary in pose, viewpoint, or background, thereby misleading the classifier. Mathematically, this framework remains consistent with traditional adversarial attacks in a principled manner. Our theoretical and empirical results demonstrate that concept-based adversarial attacks yield more diverse adversarial examples and effectively preserve the underlying concept, while achieving higher attack efficiency.

Related Topics

Attack On Titan

Massive Attack

January 6 United States Capitol Attack

Attack On Titan (Tv Series)

Joint Direct Attack Munition

2012 Benghazi Attack

Wannacry Ransomware Attack

Attack Aircraft

Shark Attack

Miami Cannibal Attack

Generative Adversarial Network

Attack On Reginald Denny

Concepts

No concepts available.

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2507.02965
PDF: https://arxiv.org/pdf/2507.02965
OA Status: green
OpenAlex ID: https://openalex.org/W4415343510

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4415343510

Canonical identifier for this work in OpenAlex
Title: Concept-based Adversarial Attack: a Probabilistic Perspective

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2025

Year of publication
Publication date: 2025-06-30

Full publication date if available
Authors: Andi Zhang, X. X. Ding, Steven McDonagh, Samuel Kaski

List of authors in order
Landing page: https://arxiv.org/abs/2507.02965

Publisher landing page
PDF URL: https://arxiv.org/pdf/2507.02965

Direct link to full text PDF
Open access: Yes

Whether a free full text is available
OA status: green

Open access status per OpenAlex
OA URL: https://arxiv.org/pdf/2507.02965

Direct OA link when available
Cited by: 0

Total citation count in OpenAlex

Full payload

id	https://openalex.org/W4415343510
doi
ids.openalex	https://openalex.org/W4415343510
fwci	0.0
type	preprint
title	Concept-based Adversarial Attack: a Probabilistic Perspective
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T11689
topics[0].field.id	https://openalex.org/fields/17
topics[0].field.display_name	Computer Science
topics[0].score	0.9818000197410583
topics[0].domain.id	https://openalex.org/domains/3
topics[0].domain.display_name	Physical Sciences
topics[0].subfield.id	https://openalex.org/subfields/1702
topics[0].subfield.display_name	Artificial Intelligence
topics[0].display_name	Adversarial Robustness in Machine Learning
is_xpac	False
apc_list
apc_paid
language	en
locations[0].id	pmh:oai:arXiv.org:2507.02965
locations[0].is_oa	True
locations[0].source.id	https://openalex.org/S4306400194
locations[0].source.issn
locations[0].source.type	repository
locations[0].source.is_oa	True
locations[0].source.issn_l
locations[0].source.is_core	False
locations[0].source.is_in_doaj	False
locations[0].source.display_name	arXiv (Cornell University)
locations[0].source.host_organization	https://openalex.org/I205783295
locations[0].source.host_organization_name	Cornell University
locations[0].source.host_organization_lineage	https://openalex.org/I205783295
locations[0].license
locations[0].pdf_url	https://arxiv.org/pdf/2507.02965
locations[0].version	submittedVersion
locations[0].raw_type	text
locations[0].license_id
locations[0].is_accepted	False
locations[0].is_published	False
locations[0].raw_source_name
locations[0].landing_page_url	http://arxiv.org/abs/2507.02965
indexed_in	arxiv
authorships[0].author.id	https://openalex.org/A5077911588
authorships[0].author.orcid	https://orcid.org/0009-0007-4855-5442
authorships[0].author.display_name	Andi Zhang
authorships[0].author_position	first
authorships[0].raw_author_name	Zhang, Andi
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5074531806
authorships[1].author.orcid	https://orcid.org/0000-0002-2807-2597
authorships[1].author.display_name	X. X. Ding
authorships[1].author_position	middle
authorships[1].raw_author_name	Ding, Xuan
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5052824649
authorships[2].author.orcid	https://orcid.org/0000-0001-7025-5197
authorships[2].author.display_name	Steven McDonagh
authorships[2].author_position	middle
authorships[2].raw_author_name	McDonagh, Steven
authorships[2].is_corresponding	False
authorships[3].author.id	https://openalex.org/A5018305257
authorships[3].author.orcid	https://orcid.org/0000-0003-1925-9154
authorships[3].author.display_name	Samuel Kaski
authorships[3].author_position	last
authorships[3].raw_author_name	Kaski, Samuel
authorships[3].is_corresponding	False
has_content.pdf	False
has_content.grobid_xml	False
is_paratext	False
open_access.is_oa	True
open_access.oa_url	https://arxiv.org/pdf/2507.02965
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-20T00:00:00
display_name	Concept-based Adversarial Attack: a Probabilistic Perspective
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T04:12:42.849631
primary_topic.id	https://openalex.org/T11689
primary_topic.field.id	https://openalex.org/fields/17
primary_topic.field.display_name	Computer Science
primary_topic.score	0.9818000197410583
primary_topic.domain.id	https://openalex.org/domains/3
primary_topic.domain.display_name	Physical Sciences
primary_topic.subfield.id	https://openalex.org/subfields/1702
primary_topic.subfield.display_name	Artificial Intelligence
primary_topic.display_name	Adversarial Robustness in Machine Learning
cited_by_count	0
locations_count	1
best_oa_location.id	pmh:oai:arXiv.org:2507.02965
best_oa_location.is_oa	True
best_oa_location.source.id	https://openalex.org/S4306400194
best_oa_location.source.issn
best_oa_location.source.type	repository
best_oa_location.source.is_oa	True
best_oa_location.source.issn_l
best_oa_location.source.is_core	False
best_oa_location.source.is_in_doaj	False
best_oa_location.source.display_name	arXiv (Cornell University)
best_oa_location.source.host_organization	https://openalex.org/I205783295
best_oa_location.source.host_organization_name	Cornell University
best_oa_location.source.host_organization_lineage	https://openalex.org/I205783295
best_oa_location.license
best_oa_location.pdf_url	https://arxiv.org/pdf/2507.02965
best_oa_location.version	submittedVersion
best_oa_location.raw_type	text
best_oa_location.license_id
best_oa_location.is_accepted	False
best_oa_location.is_published	False
best_oa_location.raw_source_name
best_oa_location.landing_page_url	http://arxiv.org/abs/2507.02965
primary_location.id	pmh:oai:arXiv.org:2507.02965
primary_location.is_oa	True
primary_location.source.id	https://openalex.org/S4306400194
primary_location.source.issn
primary_location.source.type	repository
primary_location.source.is_oa	True
primary_location.source.issn_l
primary_location.source.is_core	False
primary_location.source.is_in_doaj	False
primary_location.source.display_name	arXiv (Cornell University)
primary_location.source.host_organization	https://openalex.org/I205783295
primary_location.source.host_organization_name	Cornell University
primary_location.source.host_organization_lineage	https://openalex.org/I205783295
primary_location.license
primary_location.pdf_url	https://arxiv.org/pdf/2507.02965
primary_location.version	submittedVersion
primary_location.raw_type	text
primary_location.license_id
primary_location.is_accepted	False
primary_location.is_published	False
primary_location.raw_source_name
primary_location.landing_page_url	http://arxiv.org/abs/2507.02965
publication_date	2025-06-30
publication_year	2025
referenced_works_count	0
abstract_inverted_index.a	2, 14, 20, 33, 38, 108
abstract_inverted_index.--	30, 42
abstract_inverted_index.By	72
abstract_inverted_index.We	0
abstract_inverted_index.an	27
abstract_inverted_index.as	53, 63
abstract_inverted_index.by	12, 32
abstract_inverted_index.in	89, 107
abstract_inverted_index.is	51
abstract_inverted_index.it	54
abstract_inverted_index.of	40, 65
abstract_inverted_index.on	26
abstract_inverted_index.or	37, 70, 92
abstract_inverted_index.to	43
abstract_inverted_index.we	79
abstract_inverted_index.Our	111
abstract_inverted_index.and	113, 126
abstract_inverted_index.but	87
abstract_inverted_index.our	23
abstract_inverted_index.set	39
abstract_inverted_index.the	49, 57, 66, 84, 96, 129
abstract_inverted_index.from	74
abstract_inverted_index.more	122
abstract_inverted_index.than	18
abstract_inverted_index.that	7, 56, 82, 117
abstract_inverted_index.this	75, 99
abstract_inverted_index.vary	88
abstract_inverted_index.with	103
abstract_inverted_index.model	36
abstract_inverted_index.pose,	90
abstract_inverted_index.while	132
abstract_inverted_index.yield	121
abstract_inverted_index.Rather	17
abstract_inverted_index.attack	5, 135
abstract_inverted_index.beyond	9
abstract_inverted_index.entire	28
abstract_inverted_index.higher	134
abstract_inverted_index.image,	22
abstract_inverted_index.images	41, 60, 81
abstract_inverted_index.method	24
abstract_inverted_index.remain	61
abstract_inverted_index.single	21
abstract_inverted_index.attacks	106, 120
abstract_inverted_index.concept	29, 50, 86
abstract_inverted_index.diverse	45, 123
abstract_inverted_index.ensures	55
abstract_inverted_index.extends	8
abstract_inverted_index.manner.	110
abstract_inverted_index.propose	1
abstract_inverted_index.remains	101
abstract_inverted_index.results	115
abstract_inverted_index.thereby	94
abstract_inverted_index.adopting	13
abstract_inverted_index.category	69
abstract_inverted_index.concept,	131
abstract_inverted_index.examples	125
abstract_inverted_index.generate	44, 80
abstract_inverted_index.maintain	83
abstract_inverted_index.operates	25
abstract_inverted_index.original	67, 85
abstract_inverted_index.preserve	128
abstract_inverted_index.sampling	73
abstract_inverted_index.achieving	133
abstract_inverted_index.empirical	114
abstract_inverted_index.examples.	47
abstract_inverted_index.framework	6, 100
abstract_inverted_index.identity.	71
abstract_inverted_index.instances	64
abstract_inverted_index.modifying	19
abstract_inverted_index.resulting	58
abstract_inverted_index.Preserving	48
abstract_inverted_index.consistent	102
abstract_inverted_index.essential,	52
abstract_inverted_index.generative	35
abstract_inverted_index.misleading	95
abstract_inverted_index.principled	109
abstract_inverted_index.underlying	68, 130
abstract_inverted_index.viewpoint,	91
abstract_inverted_index.adversarial	4, 46, 59, 77, 105, 119, 124
abstract_inverted_index.background,	93
abstract_inverted_index.classifier.	97
abstract_inverted_index.demonstrate	116
abstract_inverted_index.effectively	127
abstract_inverted_index.efficiency.	136
abstract_inverted_index.represented	31
abstract_inverted_index.theoretical	112
abstract_inverted_index.traditional	104
abstract_inverted_index.identifiable	62
abstract_inverted_index.perspective.	16
abstract_inverted_index.single-image	10
abstract_inverted_index.concept-based	3, 76, 118
abstract_inverted_index.distribution,	78
abstract_inverted_index.perturbations	11
abstract_inverted_index.probabilistic	15, 34
abstract_inverted_index.Mathematically,	98
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	4
citation_normalized_percentile.value	0.22684831
citation_normalized_percentile.is_in_top_1_percent	False
citation_normalized_percentile.is_in_top_10_percent	True