Adversarial Examples, Uncertainty, and Transfer Testing Robustness in Gaussian Process Hybrid Deep Networks Article Swipe

PDF

John Bradshaw , Alexander Matthews , Zoubin Ghahramani ·

YOU? · · 2017 · Open Access · · DOI: https://doi.org/10.48550/arxiv.1707.02476

Deep neural networks (DNNs) have excellent representative power and are state of the art classifiers on many tasks. However, they often do not capture their own uncertainties well making them less robust in the real world as they overconfidently extrapolate and do not notice domain shift. Gaussian processes (GPs) with RBF kernels on the other hand have better calibrated uncertainties and do not overconfidently extrapolate far from data in their training set. However, GPs have poor representational power and do not perform as well as DNNs on complex domains. In this paper we show that GP hybrid deep networks, GPDNNs, (GPs on top of DNNs and trained end-to-end) inherit the nice properties of both GPs and DNNs and are much more robust to adversarial examples. When extrapolating to adversarial examples and testing in domain shift settings, GPDNNs frequently output high entropy class probabilities corresponding to essentially "don't know". GPDNNs are therefore promising as deep architectures that know when they don't know.

Related Topics

Deep Learning

Computer Science

Artificial Intelligence

Global Positioning System

Concepts

Deep neural networks Computer science Adversarial system Robustness (evolution) Artificial intelligence Gaussian process Global Positioning System Gaussian Machine learning Deep learning Artificial neural network Parametric statistics Mathematics Telecommunications Statistics Gene Chemistry Quantum mechanics Biochemistry Physics

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/1707.02476
PDF: https://arxiv.org/pdf/1707.02476
OA Status: green
Cited By: 122
References: 36
Related Works: 10
OpenAlex ID: https://openalex.org/W2735471653

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W2735471653

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.1707.02476

Digital Object Identifier
Title: Adversarial Examples, Uncertainty, and Transfer Testing Robustness in Gaussian Process Hybrid Deep Networks

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2017

Year of publication
Publication date: 2017-07-08

Full publication date if available
Authors: John Bradshaw, Alexander Matthews, Zoubin Ghahramani

List of authors in order
Landing page: https://arxiv.org/abs/1707.02476

Publisher landing page
PDF URL: https://arxiv.org/pdf/1707.02476

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/1707.02476

Direct OA link when available
Concepts: Deep neural networks, Computer science, Adversarial system, Robustness (evolution), Artificial intelligence, Gaussian process, Global Positioning System, Gaussian, Machine learning, Deep learning, Artificial neural network, Parametric statistics, Mathematics, Telecommunications, Statistics, Gene, Chemistry, Quantum mechanics, Biochemistry, Physics

Top concepts (fields/topics) attached by OpenAlex
Cited by: 122

Total citation count in OpenAlex
Citations by year (recent): 2024: 5, 2023: 5, 2022: 10, 2021: 31, 2020: 23

Per-year citation counts (last 5 years)
References (count): 36

Number of works referenced by this work
Related works (count): 10

Other works algorithmically related by OpenAlex

Full payload

id	https://openalex.org/W2735471653
doi	https://doi.org/10.48550/arxiv.1707.02476
ids.doi	https://doi.org/10.48550/arxiv.1707.02476
ids.mag	2735471653
ids.openalex	https://openalex.org/W2735471653
fwci
type	preprint
title	Adversarial Examples, Uncertainty, and Transfer Testing Robustness in Gaussian Process Hybrid Deep Networks
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T12814
topics[0].field.id	https://openalex.org/fields/17
topics[0].field.display_name	Computer Science
topics[0].score	0.9994000196456909
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	Gaussian Processes and Bayesian Inference
topics[1].id	https://openalex.org/T11689
topics[1].field.id	https://openalex.org/fields/17
topics[1].field.display_name	Computer Science
topics[1].score	0.9991999864578247
topics[1].domain.id	https://openalex.org/domains/3
topics[1].domain.display_name	Physical Sciences
topics[1].subfield.id	https://openalex.org/subfields/1702
topics[1].subfield.display_name	Artificial Intelligence
topics[1].display_name	Adversarial Robustness in Machine Learning
topics[2].id	https://openalex.org/T11512
topics[2].field.id	https://openalex.org/fields/17
topics[2].field.display_name	Computer Science
topics[2].score	0.9866999983787537
topics[2].domain.id	https://openalex.org/domains/3
topics[2].domain.display_name	Physical Sciences
topics[2].subfield.id	https://openalex.org/subfields/1702
topics[2].subfield.display_name	Artificial Intelligence
topics[2].display_name	Anomaly Detection Techniques and Applications
is_xpac	False
apc_list
apc_paid
concepts[0].id	https://openalex.org/C2984842247
concepts[0].level	3
concepts[0].score	0.7614560127258301
concepts[0].wikidata	https://www.wikidata.org/wiki/Q197536
concepts[0].display_name	Deep neural networks
concepts[1].id	https://openalex.org/C41008148
concepts[1].level	0
concepts[1].score	0.7074484825134277
concepts[1].wikidata	https://www.wikidata.org/wiki/Q21198
concepts[1].display_name	Computer science
concepts[2].id	https://openalex.org/C37736160
concepts[2].level	2
concepts[2].score	0.6828226447105408
concepts[2].wikidata	https://www.wikidata.org/wiki/Q1801315
concepts[2].display_name	Adversarial system
concepts[3].id	https://openalex.org/C63479239
concepts[3].level	3
concepts[3].score	0.5850481390953064
concepts[3].wikidata	https://www.wikidata.org/wiki/Q7353546
concepts[3].display_name	Robustness (evolution)
concepts[4].id	https://openalex.org/C154945302
concepts[4].level	1
concepts[4].score	0.5795367956161499
concepts[4].wikidata	https://www.wikidata.org/wiki/Q11660
concepts[4].display_name	Artificial intelligence
concepts[5].id	https://openalex.org/C61326573
concepts[5].level	3
concepts[5].score	0.5379650592803955
concepts[5].wikidata	https://www.wikidata.org/wiki/Q1496376
concepts[5].display_name	Gaussian process
concepts[6].id	https://openalex.org/C60229501
concepts[6].level	2
concepts[6].score	0.5247535705566406
concepts[6].wikidata	https://www.wikidata.org/wiki/Q18822
concepts[6].display_name	Global Positioning System
concepts[7].id	https://openalex.org/C163716315
concepts[7].level	2
concepts[7].score	0.5110777616500854
concepts[7].wikidata	https://www.wikidata.org/wiki/Q901177
concepts[7].display_name	Gaussian
concepts[8].id	https://openalex.org/C119857082
concepts[8].level	1
concepts[8].score	0.4954797029495239
concepts[8].wikidata	https://www.wikidata.org/wiki/Q2539
concepts[8].display_name	Machine learning
concepts[9].id	https://openalex.org/C108583219
concepts[9].level	2
concepts[9].score	0.46677082777023315
concepts[9].wikidata	https://www.wikidata.org/wiki/Q197536
concepts[9].display_name	Deep learning
concepts[10].id	https://openalex.org/C50644808
concepts[10].level	2
concepts[10].score	0.4471316337585449
concepts[10].wikidata	https://www.wikidata.org/wiki/Q192776
concepts[10].display_name	Artificial neural network
concepts[11].id	https://openalex.org/C117251300
concepts[11].level	2
concepts[11].score	0.43304911255836487
concepts[11].wikidata	https://www.wikidata.org/wiki/Q1849855
concepts[11].display_name	Parametric statistics
concepts[12].id	https://openalex.org/C33923547
concepts[12].level	0
concepts[12].score	0.1404629349708557
concepts[12].wikidata	https://www.wikidata.org/wiki/Q395
concepts[12].display_name	Mathematics
concepts[13].id	https://openalex.org/C76155785
concepts[13].level	1
concepts[13].score	0.1068553626537323
concepts[13].wikidata	https://www.wikidata.org/wiki/Q418
concepts[13].display_name	Telecommunications
concepts[14].id	https://openalex.org/C105795698
concepts[14].level	1
concepts[14].score	0.07827451825141907
concepts[14].wikidata	https://www.wikidata.org/wiki/Q12483
concepts[14].display_name	Statistics
concepts[15].id	https://openalex.org/C104317684
concepts[15].level	2
concepts[15].score	0.0
concepts[15].wikidata	https://www.wikidata.org/wiki/Q7187
concepts[15].display_name	Gene
concepts[16].id	https://openalex.org/C185592680
concepts[16].level	0
concepts[16].score	0.0
concepts[16].wikidata	https://www.wikidata.org/wiki/Q2329
concepts[16].display_name	Chemistry
concepts[17].id	https://openalex.org/C62520636
concepts[17].level	1
concepts[17].score	0.0
concepts[17].wikidata	https://www.wikidata.org/wiki/Q944
concepts[17].display_name	Quantum mechanics
concepts[18].id	https://openalex.org/C55493867
concepts[18].level	1
concepts[18].score	0.0
concepts[18].wikidata	https://www.wikidata.org/wiki/Q7094
concepts[18].display_name	Biochemistry
concepts[19].id	https://openalex.org/C121332964
concepts[19].level	0
concepts[19].score	0.0
concepts[19].wikidata	https://www.wikidata.org/wiki/Q413
concepts[19].display_name	Physics
keywords[0].id	https://openalex.org/keywords/deep-neural-networks
keywords[0].score	0.7614560127258301
keywords[0].display_name	Deep neural networks
keywords[1].id	https://openalex.org/keywords/computer-science
keywords[1].score	0.7074484825134277
keywords[1].display_name	Computer science
keywords[2].id	https://openalex.org/keywords/adversarial-system
keywords[2].score	0.6828226447105408
keywords[2].display_name	Adversarial system
keywords[3].id	https://openalex.org/keywords/robustness
keywords[3].score	0.5850481390953064
keywords[3].display_name	Robustness (evolution)
keywords[4].id	https://openalex.org/keywords/artificial-intelligence
keywords[4].score	0.5795367956161499
keywords[4].display_name	Artificial intelligence
keywords[5].id	https://openalex.org/keywords/gaussian-process
keywords[5].score	0.5379650592803955
keywords[5].display_name	Gaussian process
keywords[6].id	https://openalex.org/keywords/global-positioning-system
keywords[6].score	0.5247535705566406
keywords[6].display_name	Global Positioning System
keywords[7].id	https://openalex.org/keywords/gaussian
keywords[7].score	0.5110777616500854
keywords[7].display_name	Gaussian
keywords[8].id	https://openalex.org/keywords/machine-learning
keywords[8].score	0.4954797029495239
keywords[8].display_name	Machine learning
keywords[9].id	https://openalex.org/keywords/deep-learning
keywords[9].score	0.46677082777023315
keywords[9].display_name	Deep learning
keywords[10].id	https://openalex.org/keywords/artificial-neural-network
keywords[10].score	0.4471316337585449
keywords[10].display_name	Artificial neural network
keywords[11].id	https://openalex.org/keywords/parametric-statistics
keywords[11].score	0.43304911255836487
keywords[11].display_name	Parametric statistics
keywords[12].id	https://openalex.org/keywords/mathematics
keywords[12].score	0.1404629349708557
keywords[12].display_name	Mathematics
keywords[13].id	https://openalex.org/keywords/telecommunications
keywords[13].score	0.1068553626537323
keywords[13].display_name	Telecommunications
keywords[14].id	https://openalex.org/keywords/statistics
keywords[14].score	0.07827451825141907
keywords[14].display_name	Statistics
language	en
locations[0].id	pmh:oai:arXiv.org:1707.02476
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/1707.02476
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/1707.02476
locations[1].id	doi:10.48550/arxiv.1707.02476
locations[1].is_oa	True
locations[1].source.id	https://openalex.org/S4306400194
locations[1].source.issn
locations[1].source.type	repository
locations[1].source.is_oa	True
locations[1].source.issn_l
locations[1].source.is_core	False
locations[1].source.is_in_doaj	False
locations[1].source.display_name	arXiv (Cornell University)
locations[1].source.host_organization	https://openalex.org/I205783295
locations[1].source.host_organization_name	Cornell University
locations[1].source.host_organization_lineage	https://openalex.org/I205783295
locations[1].license
locations[1].pdf_url
locations[1].version
locations[1].raw_type	article
locations[1].license_id
locations[1].is_accepted	False
locations[1].is_published
locations[1].raw_source_name
locations[1].landing_page_url	https://doi.org/10.48550/arxiv.1707.02476
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5112088138
authorships[0].author.orcid
authorships[0].author.display_name	John Bradshaw
authorships[0].author_position	first
authorships[0].raw_author_name	John Bradshaw
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5042500383
authorships[1].author.orcid	https://orcid.org/0000-0002-7279-9726
authorships[1].author.display_name	Alexander Matthews
authorships[1].author_position	middle
authorships[1].raw_author_name	Alexander G. de G. Matthews
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5010820865
authorships[2].author.orcid	https://orcid.org/0000-0002-7464-6475
authorships[2].author.display_name	Zoubin Ghahramani
authorships[2].author_position	last
authorships[2].raw_author_name	Zoubin Ghahramani
authorships[2].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/1707.02476
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-10T00:00:00
display_name	Adversarial Examples, Uncertainty, and Transfer Testing Robustness in Gaussian Process Hybrid Deep Networks
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T06:51:31.235846
primary_topic.id	https://openalex.org/T12814
primary_topic.field.id	https://openalex.org/fields/17
primary_topic.field.display_name	Computer Science
primary_topic.score	0.9994000196456909
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	Gaussian Processes and Bayesian Inference
related_works	https://openalex.org/W2950183588, https://openalex.org/W3080754722, https://openalex.org/W4383221314, https://openalex.org/W3093978547, https://openalex.org/W2953536436, https://openalex.org/W3203790781, https://openalex.org/W4313346231, https://openalex.org/W2738001131, https://openalex.org/W4285785480, https://openalex.org/W2997056298
cited_by_count	122
counts_by_year[0].year	2024
counts_by_year[0].cited_by_count	5
counts_by_year[1].year	2023
counts_by_year[1].cited_by_count	5
counts_by_year[2].year	2022
counts_by_year[2].cited_by_count	10
counts_by_year[3].year	2021
counts_by_year[3].cited_by_count	31
counts_by_year[4].year	2020
counts_by_year[4].cited_by_count	23
counts_by_year[5].year	2019
counts_by_year[5].cited_by_count	18
counts_by_year[6].year	2018
counts_by_year[6].cited_by_count	25
counts_by_year[7].year	2017
counts_by_year[7].cited_by_count	5
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:1707.02476
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/1707.02476
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/1707.02476
primary_location.id	pmh:oai:arXiv.org:1707.02476
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/1707.02476
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/1707.02476
publication_date	2017-07-08
publication_year	2017
referenced_works	https://openalex.org/W2342840547, https://openalex.org/W2511730936, https://openalex.org/W2335728318, https://openalex.org/W2605055943, https://openalex.org/W2953047670, https://openalex.org/W2163605009, https://openalex.org/W2162724919, https://openalex.org/W2963651065, https://openalex.org/W2952677397, https://openalex.org/W2243397390, https://openalex.org/W2570685808, https://openalex.org/W2103328396, https://openalex.org/W2603766943, https://openalex.org/W2964153729, https://openalex.org/W1945616565, https://openalex.org/W2271840356, https://openalex.org/W2137135057, https://openalex.org/W3118608800, https://openalex.org/W2551835155, https://openalex.org/W2108677974, https://openalex.org/W2408141691, https://openalex.org/W2427628320, https://openalex.org/W2160815625, https://openalex.org/W2516574342, https://openalex.org/W2612637113, https://openalex.org/W1966948031, https://openalex.org/W2149991584, https://openalex.org/W2460937040, https://openalex.org/W2148319469, https://openalex.org/W2963826681, https://openalex.org/W2951308445, https://openalex.org/W2728139190, https://openalex.org/W2951985269, https://openalex.org/W2964135722, https://openalex.org/W2154579312, https://openalex.org/W2810611310
referenced_works_count	36
abstract_inverted_index.GP	95
abstract_inverted_index.In	89
abstract_inverted_index.as	36, 82, 84, 152
abstract_inverted_index.do	21, 41, 61, 79
abstract_inverted_index.in	32, 68, 132
abstract_inverted_index.of	11, 103, 112
abstract_inverted_index.on	15, 52, 86, 101
abstract_inverted_index.to	122, 127, 144
abstract_inverted_index.we	92
abstract_inverted_index.GPs	73, 114
abstract_inverted_index.RBF	50
abstract_inverted_index.and	8, 40, 60, 78, 105, 115, 117, 130
abstract_inverted_index.are	9, 118, 149
abstract_inverted_index.art	13
abstract_inverted_index.far	65
abstract_inverted_index.not	22, 42, 62, 80
abstract_inverted_index.own	25
abstract_inverted_index.the	12, 33, 53, 109
abstract_inverted_index.top	102
abstract_inverted_index.(GPs	100
abstract_inverted_index.DNNs	85, 104, 116
abstract_inverted_index.Deep	0
abstract_inverted_index.When	125
abstract_inverted_index.both	113
abstract_inverted_index.data	67
abstract_inverted_index.deep	97, 153
abstract_inverted_index.from	66
abstract_inverted_index.hand	55
abstract_inverted_index.have	4, 56, 74
abstract_inverted_index.high	139
abstract_inverted_index.know	156
abstract_inverted_index.less	30
abstract_inverted_index.many	16
abstract_inverted_index.more	120
abstract_inverted_index.much	119
abstract_inverted_index.nice	110
abstract_inverted_index.poor	75
abstract_inverted_index.real	34
abstract_inverted_index.set.	71
abstract_inverted_index.show	93
abstract_inverted_index.that	94, 155
abstract_inverted_index.them	29
abstract_inverted_index.they	19, 37, 158
abstract_inverted_index.this	90
abstract_inverted_index.well	27, 83
abstract_inverted_index.when	157
abstract_inverted_index.with	49
abstract_inverted_index.(GPs)	48
abstract_inverted_index.class	141
abstract_inverted_index.don't	159
abstract_inverted_index.know.	160
abstract_inverted_index.often	20
abstract_inverted_index.other	54
abstract_inverted_index.paper	91
abstract_inverted_index.power	7, 77
abstract_inverted_index.shift	134
abstract_inverted_index.state	10
abstract_inverted_index.their	24, 69
abstract_inverted_index.world	35
abstract_inverted_index."don't	146
abstract_inverted_index.(DNNs)	3
abstract_inverted_index.GPDNNs	136, 148
abstract_inverted_index.better	57
abstract_inverted_index.domain	44, 133
abstract_inverted_index.hybrid	96
abstract_inverted_index.know".	147
abstract_inverted_index.making	28
abstract_inverted_index.neural	1
abstract_inverted_index.notice	43
abstract_inverted_index.output	138
abstract_inverted_index.robust	31, 121
abstract_inverted_index.shift.	45
abstract_inverted_index.tasks.	17
abstract_inverted_index.GPDNNs,	99
abstract_inverted_index.capture	23
abstract_inverted_index.complex	87
abstract_inverted_index.entropy	140
abstract_inverted_index.inherit	108
abstract_inverted_index.kernels	51
abstract_inverted_index.perform	81
abstract_inverted_index.testing	131
abstract_inverted_index.trained	106
abstract_inverted_index.Gaussian	46
abstract_inverted_index.However,	18, 72
abstract_inverted_index.domains.	88
abstract_inverted_index.examples	129
abstract_inverted_index.networks	2
abstract_inverted_index.training	70
abstract_inverted_index.examples.	124
abstract_inverted_index.excellent	5
abstract_inverted_index.networks,	98
abstract_inverted_index.processes	47
abstract_inverted_index.promising	151
abstract_inverted_index.settings,	135
abstract_inverted_index.therefore	150
abstract_inverted_index.calibrated	58
abstract_inverted_index.frequently	137
abstract_inverted_index.properties	111
abstract_inverted_index.adversarial	123, 128
abstract_inverted_index.classifiers	14
abstract_inverted_index.end-to-end)	107
abstract_inverted_index.essentially	145
abstract_inverted_index.extrapolate	39, 64
abstract_inverted_index.architectures	154
abstract_inverted_index.corresponding	143
abstract_inverted_index.extrapolating	126
abstract_inverted_index.probabilities	142
abstract_inverted_index.uncertainties	26, 59
abstract_inverted_index.representative	6
abstract_inverted_index.overconfidently	38, 63
abstract_inverted_index.representational	76
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	3
citation_normalized_percentile