VQ4ALL: Efficient Neural Network Representation via a Universal Codebook Article Swipe

PDF

Juncan Deng , Shuaiting Li , Zeyu Wang , Hong Gu , Kedong Xu , Kejie Huang ·

YOU? · · 2024 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2412.06875

The rapid growth of the big neural network models puts forward new requirements for lightweight network representation methods. The traditional methods based on model compression have achieved great success, especially VQ technology which realizes the high compression ratio of models by sharing code words. However, because each layer of the network needs to build a code table, the traditional top-down compression technology lacks attention to the underlying commonalities, resulting in limited compression rate and frequent memory access. In this paper, we propose a bottom-up method to share the universal codebook among multiple neural networks, which not only effectively reduces the number of codebooks but also further reduces the memory access and chip area by storing static code tables in the built-in ROM. Specifically, we introduce VQ4ALL, a VQ-based method that utilizes codewords to enable the construction of various neural networks and achieve efficient representations. The core idea of our method is to adopt a kernel density estimation approach to extract a universal codebook and then progressively construct different low-bit networks by updating differentiable assignments. Experimental results demonstrate that VQ4ALL achieves compression rates exceeding 16 $\times$ while preserving high accuracy across multiple network architectures, highlighting its effectiveness and versatility.

Related Topics

Computer Science

Artificial Intelligence

Politics

Law

Concepts

Codebook Representation (politics) Artificial neural network Computer science Artificial intelligence Political science Politics Law

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2412.06875
PDF: https://arxiv.org/pdf/2412.06875
OA Status: green
Related Works: 10
OpenAlex ID: https://openalex.org/W4405254655

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4405254655

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.2412.06875

Digital Object Identifier
Title: VQ4ALL: Efficient Neural Network Representation via a Universal Codebook

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2024

Year of publication
Publication date: 2024-12-09

Full publication date if available
Authors: Juncan Deng, Shuaiting Li, Zeyu Wang, Hong Gu, Kedong Xu, Kejie Huang

List of authors in order
Landing page: https://arxiv.org/abs/2412.06875

Publisher landing page
PDF URL: https://arxiv.org/pdf/2412.06875

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/2412.06875

Direct OA link when available
Concepts: Codebook, Representation (politics), Artificial neural network, Computer science, Artificial intelligence, Political science, Politics, Law

Top concepts (fields/topics) attached by OpenAlex
Cited by: 0

Total citation count in OpenAlex
Related works (count): 10

Other works algorithmically related by OpenAlex

Full payload

id	https://openalex.org/W4405254655
doi	https://doi.org/10.48550/arxiv.2412.06875
ids.doi	https://doi.org/10.48550/arxiv.2412.06875
ids.openalex	https://openalex.org/W4405254655
fwci
type	preprint
title	VQ4ALL: Efficient Neural Network Representation via a Universal Codebook
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T10320
topics[0].field.id	https://openalex.org/fields/17
topics[0].field.display_name	Computer Science
topics[0].score	0.8299999833106995
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	Neural Networks and Applications
is_xpac	False
apc_list
apc_paid
concepts[0].id	https://openalex.org/C127759330
concepts[0].level	2
concepts[0].score	0.9469667673110962
concepts[0].wikidata	https://www.wikidata.org/wiki/Q637416
concepts[0].display_name	Codebook
concepts[1].id	https://openalex.org/C2776359362
concepts[1].level	3
concepts[1].score	0.661614179611206
concepts[1].wikidata	https://www.wikidata.org/wiki/Q2145286
concepts[1].display_name	Representation (politics)
concepts[2].id	https://openalex.org/C50644808
concepts[2].level	2
concepts[2].score	0.5872888565063477
concepts[2].wikidata	https://www.wikidata.org/wiki/Q192776
concepts[2].display_name	Artificial neural network
concepts[3].id	https://openalex.org/C41008148
concepts[3].level	0
concepts[3].score	0.5338541269302368
concepts[3].wikidata	https://www.wikidata.org/wiki/Q21198
concepts[3].display_name	Computer science
concepts[4].id	https://openalex.org/C154945302
concepts[4].level	1
concepts[4].score	0.3681481182575226
concepts[4].wikidata	https://www.wikidata.org/wiki/Q11660
concepts[4].display_name	Artificial intelligence
concepts[5].id	https://openalex.org/C17744445
concepts[5].level	0
concepts[5].score	0.04896506667137146
concepts[5].wikidata	https://www.wikidata.org/wiki/Q36442
concepts[5].display_name	Political science
concepts[6].id	https://openalex.org/C94625758
concepts[6].level	2
concepts[6].score	0.0
concepts[6].wikidata	https://www.wikidata.org/wiki/Q7163
concepts[6].display_name	Politics
concepts[7].id	https://openalex.org/C199539241
concepts[7].level	1
concepts[7].score	0.0
concepts[7].wikidata	https://www.wikidata.org/wiki/Q7748
concepts[7].display_name	Law
keywords[0].id	https://openalex.org/keywords/codebook
keywords[0].score	0.9469667673110962
keywords[0].display_name	Codebook
keywords[1].id	https://openalex.org/keywords/representation
keywords[1].score	0.661614179611206
keywords[1].display_name	Representation (politics)
keywords[2].id	https://openalex.org/keywords/artificial-neural-network
keywords[2].score	0.5872888565063477
keywords[2].display_name	Artificial neural network
keywords[3].id	https://openalex.org/keywords/computer-science
keywords[3].score	0.5338541269302368
keywords[3].display_name	Computer science
keywords[4].id	https://openalex.org/keywords/artificial-intelligence
keywords[4].score	0.3681481182575226
keywords[4].display_name	Artificial intelligence
keywords[5].id	https://openalex.org/keywords/political-science
keywords[5].score	0.04896506667137146
keywords[5].display_name	Political science
language	en
locations[0].id	pmh:oai:arXiv.org:2412.06875
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/2412.06875
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/2412.06875
locations[1].id	doi:10.48550/arxiv.2412.06875
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	cc-by
locations[1].pdf_url
locations[1].version
locations[1].raw_type	article
locations[1].license_id	https://openalex.org/licenses/cc-by
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.2412.06875
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5065813707
authorships[0].author.orcid	https://orcid.org/0000-0003-0860-4442
authorships[0].author.display_name	Juncan Deng
authorships[0].author_position	first
authorships[0].raw_author_name	Deng, Juncan
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5019049217
authorships[1].author.orcid	https://orcid.org/0009-0002-7726-4883
authorships[1].author.display_name	Shuaiting Li
authorships[1].author_position	middle
authorships[1].raw_author_name	Li, Shuaiting
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5100379682
authorships[2].author.orcid	https://orcid.org/0000-0001-5374-6330
authorships[2].author.display_name	Zeyu Wang
authorships[2].author_position	middle
authorships[2].raw_author_name	Wang, Zeyu
authorships[2].is_corresponding	False
authorships[3].author.id	https://openalex.org/A5101434406
authorships[3].author.orcid	https://orcid.org/0000-0002-8224-146X
authorships[3].author.display_name	Hong Gu
authorships[3].author_position	middle
authorships[3].raw_author_name	Gu, Hong
authorships[3].is_corresponding	False
authorships[4].author.id	https://openalex.org/A5111345577
authorships[4].author.orcid
authorships[4].author.display_name	Kedong Xu
authorships[4].author_position	middle
authorships[4].raw_author_name	Xu, Kedong
authorships[4].is_corresponding	False
authorships[5].author.id	https://openalex.org/A5091073331
authorships[5].author.orcid	https://orcid.org/0000-0003-3722-9979
authorships[5].author.display_name	Kejie Huang
authorships[5].author_position	last
authorships[5].raw_author_name	Huang, Kejie
authorships[5].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/2412.06875
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-10T00:00:00
display_name	VQ4ALL: Efficient Neural Network Representation via a Universal Codebook
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T06:51:31.235846
primary_topic.id	https://openalex.org/T10320
primary_topic.field.id	https://openalex.org/fields/17
primary_topic.field.display_name	Computer Science
primary_topic.score	0.8299999833106995
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	Neural Networks and Applications
related_works	https://openalex.org/W4391375266, https://openalex.org/W2899084033, https://openalex.org/W2748952813, https://openalex.org/W2293149949, https://openalex.org/W2026099691, https://openalex.org/W4284672201, https://openalex.org/W2377486419, https://openalex.org/W2943202426, https://openalex.org/W2137816434, https://openalex.org/W2163679795
cited_by_count	0
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2412.06875
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/2412.06875
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/2412.06875
primary_location.id	pmh:oai:arXiv.org:2412.06875
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/2412.06875
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/2412.06875
publication_date	2024-12-09
publication_year	2024
referenced_works_count	0
abstract_inverted_index.a	54, 82, 126, 153, 160
abstract_inverted_index.16	183
abstract_inverted_index.In	77
abstract_inverted_index.VQ	30
abstract_inverted_index.by	40, 113, 170
abstract_inverted_index.in	69, 118
abstract_inverted_index.is	150
abstract_inverted_index.of	3, 38, 48, 101, 136, 147
abstract_inverted_index.on	22
abstract_inverted_index.to	52, 64, 85, 132, 151, 158
abstract_inverted_index.we	80, 123
abstract_inverted_index.The	0, 18, 144
abstract_inverted_index.and	73, 110, 140, 163, 196
abstract_inverted_index.big	5
abstract_inverted_index.but	103
abstract_inverted_index.for	13
abstract_inverted_index.its	194
abstract_inverted_index.new	11
abstract_inverted_index.not	95
abstract_inverted_index.our	148
abstract_inverted_index.the	4, 34, 49, 57, 65, 87, 99, 107, 119, 134
abstract_inverted_index.ROM.	121
abstract_inverted_index.also	104
abstract_inverted_index.area	112
abstract_inverted_index.chip	111
abstract_inverted_index.code	42, 55, 116
abstract_inverted_index.core	145
abstract_inverted_index.each	46
abstract_inverted_index.have	25
abstract_inverted_index.high	35, 187
abstract_inverted_index.idea	146
abstract_inverted_index.only	96
abstract_inverted_index.puts	9
abstract_inverted_index.rate	72
abstract_inverted_index.that	129, 177
abstract_inverted_index.then	164
abstract_inverted_index.this	78
abstract_inverted_index.adopt	152
abstract_inverted_index.among	90
abstract_inverted_index.based	21
abstract_inverted_index.build	53
abstract_inverted_index.great	27
abstract_inverted_index.lacks	62
abstract_inverted_index.layer	47
abstract_inverted_index.model	23
abstract_inverted_index.needs	51
abstract_inverted_index.rapid	1
abstract_inverted_index.rates	181
abstract_inverted_index.ratio	37
abstract_inverted_index.share	86
abstract_inverted_index.which	32, 94
abstract_inverted_index.while	185
abstract_inverted_index.VQ4ALL	178
abstract_inverted_index.access	109
abstract_inverted_index.across	189
abstract_inverted_index.enable	133
abstract_inverted_index.growth	2
abstract_inverted_index.kernel	154
abstract_inverted_index.memory	75, 108
abstract_inverted_index.method	84, 128, 149
abstract_inverted_index.models	8, 39
abstract_inverted_index.neural	6, 92, 138
abstract_inverted_index.number	100
abstract_inverted_index.paper,	79
abstract_inverted_index.static	115
abstract_inverted_index.table,	56
abstract_inverted_index.tables	117
abstract_inverted_index.words.	43
abstract_inverted_index.VQ4ALL,	125
abstract_inverted_index.access.	76
abstract_inverted_index.achieve	141
abstract_inverted_index.because	45
abstract_inverted_index.density	155
abstract_inverted_index.extract	159
abstract_inverted_index.forward	10
abstract_inverted_index.further	105
abstract_inverted_index.limited	70
abstract_inverted_index.low-bit	168
abstract_inverted_index.methods	20
abstract_inverted_index.network	7, 15, 50, 191
abstract_inverted_index.propose	81
abstract_inverted_index.reduces	98, 106
abstract_inverted_index.results	175
abstract_inverted_index.sharing	41
abstract_inverted_index.storing	114
abstract_inverted_index.various	137
abstract_inverted_index.$\times$	184
abstract_inverted_index.However,	44
abstract_inverted_index.VQ-based	127
abstract_inverted_index.accuracy	188
abstract_inverted_index.achieved	26
abstract_inverted_index.achieves	179
abstract_inverted_index.approach	157
abstract_inverted_index.built-in	120
abstract_inverted_index.codebook	89, 162
abstract_inverted_index.frequent	74
abstract_inverted_index.methods.	17
abstract_inverted_index.multiple	91, 190
abstract_inverted_index.networks	139, 169
abstract_inverted_index.realizes	33
abstract_inverted_index.success,	28
abstract_inverted_index.top-down	59
abstract_inverted_index.updating	171
abstract_inverted_index.utilizes	130
abstract_inverted_index.attention	63
abstract_inverted_index.bottom-up	83
abstract_inverted_index.codebooks	102
abstract_inverted_index.codewords	131
abstract_inverted_index.construct	166
abstract_inverted_index.different	167
abstract_inverted_index.efficient	142
abstract_inverted_index.exceeding	182
abstract_inverted_index.introduce	124
abstract_inverted_index.networks,	93
abstract_inverted_index.resulting	68
abstract_inverted_index.universal	88, 161
abstract_inverted_index.especially	29
abstract_inverted_index.estimation	156
abstract_inverted_index.preserving	186
abstract_inverted_index.technology	31, 61
abstract_inverted_index.underlying	66
abstract_inverted_index.compression	24, 36, 60, 71, 180
abstract_inverted_index.demonstrate	176
abstract_inverted_index.effectively	97
abstract_inverted_index.lightweight	14
abstract_inverted_index.traditional	19, 58
abstract_inverted_index.Experimental	174
abstract_inverted_index.assignments.	173
abstract_inverted_index.construction	135
abstract_inverted_index.highlighting	193
abstract_inverted_index.requirements	12
abstract_inverted_index.versatility.	197
abstract_inverted_index.Specifically,	122
abstract_inverted_index.effectiveness	195
abstract_inverted_index.progressively	165
abstract_inverted_index.architectures,	192
abstract_inverted_index.commonalities,	67
abstract_inverted_index.differentiable	172
abstract_inverted_index.representation	16
abstract_inverted_index.representations.	143
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	6
citation_normalized_percentile