Deterministic Autoencoder using Wasserstein loss for tabular data generation

Tabular data generation is a complex task due to its distinctive characteristics and inherent complexities. While Variational Autoencoders have been adapted from the computer vision domain for tabular data synthesis, their reliance on non-deterministic latent space regularization introduces limitati...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Neural networks Jg. 185; S. 107208
Hauptverfasser: Wang, Alex X., Nguyen, Binh P.
Format: Journal Article
Sprache:Englisch
Veröffentlicht: United States Elsevier Ltd 01.05.2025
Schlagworte:
ISSN:0893-6080, 1879-2782, 1879-2782
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract Tabular data generation is a complex task due to its distinctive characteristics and inherent complexities. While Variational Autoencoders have been adapted from the computer vision domain for tabular data synthesis, their reliance on non-deterministic latent space regularization introduces limitations. The stochastic nature of Variational Autoencoders can contribute to collapsed posteriors, yielding suboptimal outcomes and limiting control over the latent space. This characteristic also constrains the exploration of latent space interpolation. To address these challenges, we present the Tabular Wasserstein Autoencoder (TWAE), leveraging the deterministic encoding mechanism of Wasserstein Autoencoders. This characteristic facilitates a deterministic mapping of inputs to latent codes, enhancing the stability and expressiveness of our model’s latent space. This, in turn, enables seamless integration with shallow interpolation mechanisms like the synthetic minority over-sampling technique (SMOTE) within the data generation process via deep learning. Specifically, TWAE is trained once to establish a low-dimensional representation of real data, and various latent interpolation methods efficiently generate synthetic latent points, achieving a balance between accuracy and efficiency. Extensive experiments consistently demonstrate TWAE’s superiority, showcasing its versatility across diverse feature types and dataset sizes. This innovative approach, combining WAE principles with shallow interpolation, effectively leverages SMOTE’s advantages, establishing TWAE as a robust solution for complex tabular data synthesis.
AbstractList Tabular data generation is a complex task due to its distinctive characteristics and inherent complexities. While Variational Autoencoders have been adapted from the computer vision domain for tabular data synthesis, their reliance on non-deterministic latent space regularization introduces limitations. The stochastic nature of Variational Autoencoders can contribute to collapsed posteriors, yielding suboptimal outcomes and limiting control over the latent space. This characteristic also constrains the exploration of latent space interpolation. To address these challenges, we present the Tabular Wasserstein Autoencoder (TWAE), leveraging the deterministic encoding mechanism of Wasserstein Autoencoders. This characteristic facilitates a deterministic mapping of inputs to latent codes, enhancing the stability and expressiveness of our model’s latent space. This, in turn, enables seamless integration with shallow interpolation mechanisms like the synthetic minority over-sampling technique (SMOTE) within the data generation process via deep learning. Specifically, TWAE is trained once to establish a low-dimensional representation of real data, and various latent interpolation methods efficiently generate synthetic latent points, achieving a balance between accuracy and efficiency. Extensive experiments consistently demonstrate TWAE’s superiority, showcasing its versatility across diverse feature types and dataset sizes. This innovative approach, combining WAE principles with shallow interpolation, effectively leverages SMOTE’s advantages, establishing TWAE as a robust solution for complex tabular data synthesis.
Tabular data generation is a complex task due to its distinctive characteristics and inherent complexities. While Variational Autoencoders have been adapted from the computer vision domain for tabular data synthesis, their reliance on non-deterministic latent space regularization introduces limitations. The stochastic nature of Variational Autoencoders can contribute to collapsed posteriors, yielding suboptimal outcomes and limiting control over the latent space. This characteristic also constrains the exploration of latent space interpolation. To address these challenges, we present the Tabular Wasserstein Autoencoder (TWAE), leveraging the deterministic encoding mechanism of Wasserstein Autoencoders. This characteristic facilitates a deterministic mapping of inputs to latent codes, enhancing the stability and expressiveness of our model's latent space. This, in turn, enables seamless integration with shallow interpolation mechanisms like the synthetic minority over-sampling technique (SMOTE) within the data generation process via deep learning. Specifically, TWAE is trained once to establish a low-dimensional representation of real data, and various latent interpolation methods efficiently generate synthetic latent points, achieving a balance between accuracy and efficiency. Extensive experiments consistently demonstrate TWAE's superiority, showcasing its versatility across diverse feature types and dataset sizes. This innovative approach, combining WAE principles with shallow interpolation, effectively leverages SMOTE's advantages, establishing TWAE as a robust solution for complex tabular data synthesis.Tabular data generation is a complex task due to its distinctive characteristics and inherent complexities. While Variational Autoencoders have been adapted from the computer vision domain for tabular data synthesis, their reliance on non-deterministic latent space regularization introduces limitations. The stochastic nature of Variational Autoencoders can contribute to collapsed posteriors, yielding suboptimal outcomes and limiting control over the latent space. This characteristic also constrains the exploration of latent space interpolation. To address these challenges, we present the Tabular Wasserstein Autoencoder (TWAE), leveraging the deterministic encoding mechanism of Wasserstein Autoencoders. This characteristic facilitates a deterministic mapping of inputs to latent codes, enhancing the stability and expressiveness of our model's latent space. This, in turn, enables seamless integration with shallow interpolation mechanisms like the synthetic minority over-sampling technique (SMOTE) within the data generation process via deep learning. Specifically, TWAE is trained once to establish a low-dimensional representation of real data, and various latent interpolation methods efficiently generate synthetic latent points, achieving a balance between accuracy and efficiency. Extensive experiments consistently demonstrate TWAE's superiority, showcasing its versatility across diverse feature types and dataset sizes. This innovative approach, combining WAE principles with shallow interpolation, effectively leverages SMOTE's advantages, establishing TWAE as a robust solution for complex tabular data synthesis.
ArticleNumber 107208
Author Nguyen, Binh P.
Wang, Alex X.
Author_xml – sequence: 1
  givenname: Alex X.
  orcidid: 0000-0002-3691-8652
  surname: Wang
  fullname: Wang, Alex X.
  email: alex.wang@vuw.ac.nz
  organization: School of Mathematics and Statistics, Victoria University of Wellington, Wellington 6012, New Zealand
– sequence: 2
  givenname: Binh P.
  orcidid: 0000-0001-6203-6664
  surname: Nguyen
  fullname: Nguyen, Binh P.
  email: binh.p.nguyen@vuw.ac.nz
  organization: School of Mathematics and Statistics, Victoria University of Wellington, Wellington 6012, New Zealand
BackLink https://www.ncbi.nlm.nih.gov/pubmed/39893805$$D View this record in MEDLINE/PubMed
BookMark eNqFkMtKAzEUhoMo9qJvIDJLN1OTzC3jQij1CgU3isuQSc6UlGlSk4zg25s67caFrg6c8_0_nG-Cjo01gNAFwTOCSXm9nhnoDYQZxbSIq4pidoTGhFV1SitGj9EYszpLS8zwCE28X2OMS5Znp2iU1fHCcDFGyzsI4DbaaB-0TOZ9sGCkVeCS3muzSt6F9-B8AG2SznqftNYlQTR9J1yiRBDJCgw4EbQ1Z-ikFZ2H8_2coreH-9fFU7p8eXxezJepzDELaaEakNCITIBi0AqiamBKKJFRBg0IVlLW1kUOSjUKFzVuZVXKtqZlJZumJtkUXQ29W2c_evCBb7SX0HXCgO09z0hsyEuS5xG93KN9swHFt05vhPviBwMRuBkA6eJ3Dloudfj5JjihO04w3-nmaz7o5jvdfNAdw_mv8KH_n9jtEIMo6VOD417qqB2UdiADV1b_XfAN6kGeNw
CitedBy_id crossref_primary_10_7717_peerj_cs_2732
crossref_primary_10_1016_j_engappai_2025_110819
crossref_primary_10_3390_electronics14112230
crossref_primary_10_1007_s10115_025_02377_7
crossref_primary_10_1038_s41598_025_05545_5
Cites_doi 10.1016/j.asoc.2024.112223
10.1186/s40537-023-00792-7
10.1145/3534678.3539454
10.1609/aaai.v33i01.33015049
10.1007/s44163-021-00016-y
10.1613/jair.953
10.1016/j.asoc.2023.110895
10.1016/j.ins.2021.12.018
10.1145/3583780.3615202
10.1609/aaai.v33i01.33014610
10.29012/jpc.v1i1.568
10.1016/j.ins.2024.121610
10.1146/annurev-statistics-040720-031848
10.1016/j.inffus.2021.11.011
10.1038/s41467-022-35295-1
10.1016/j.ins.2023.02.004
10.1016/j.ipm.2023.103558
10.1109/CVPR42600.2020.00611
ContentType Journal Article
Copyright 2025 The Authors
Copyright © 2025 The Authors. Published by Elsevier Ltd.. All rights reserved.
Copyright_xml – notice: 2025 The Authors
– notice: Copyright © 2025 The Authors. Published by Elsevier Ltd.. All rights reserved.
DBID 6I.
AAFTH
AAYXX
CITATION
CGR
CUY
CVF
ECM
EIF
NPM
7X8
DOI 10.1016/j.neunet.2025.107208
DatabaseName ScienceDirect Open Access Titles
Elsevier:ScienceDirect:Open Access
CrossRef
Medline
MEDLINE
MEDLINE (Ovid)
MEDLINE
MEDLINE
PubMed
MEDLINE - Academic
DatabaseTitle CrossRef
MEDLINE
Medline Complete
MEDLINE with Full Text
PubMed
MEDLINE (Ovid)
MEDLINE - Academic
DatabaseTitleList
MEDLINE - Academic
MEDLINE
Database_xml – sequence: 1
  dbid: NPM
  name: PubMed
  url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed
  sourceTypes: Index Database
– sequence: 2
  dbid: 7X8
  name: MEDLINE - Academic
  url: https://search.proquest.com/medline
  sourceTypes: Aggregation Database
DeliveryMethod fulltext_linktorsrc
Discipline Computer Science
EISSN 1879-2782
ExternalDocumentID 39893805
10_1016_j_neunet_2025_107208
S0893608025000875
Genre Journal Article
GroupedDBID ---
--K
--M
-~X
.DC
.~1
0R~
123
186
1B1
1RT
1~.
1~5
29N
4.4
457
4G.
53G
5RE
5VS
6I.
6TJ
7-5
71M
8P~
9JM
9JN
AABNK
AAEDT
AAEDW
AAFTH
AAIKJ
AAKOC
AALRI
AAOAW
AAQFI
AAQXK
AATTM
AAXKI
AAXLA
AAXUO
AAYFN
AAYWO
ABAOU
ABBOA
ABCQJ
ABDPE
ABEFU
ABFNM
ABFRF
ABHFT
ABIVO
ABJNI
ABLJU
ABMAC
ABWVN
ABXDB
ACDAQ
ACGFO
ACGFS
ACIUM
ACNNM
ACRLP
ACRPL
ACVFH
ACZNC
ADBBV
ADCNI
ADEZE
ADGUI
ADJOM
ADMUD
ADNMO
ADRHT
AEBSH
AECPX
AEFWE
AEIPS
AEKER
AENEX
AEUPX
AFJKZ
AFPUW
AFTJW
AFXIZ
AGCQF
AGHFR
AGQPQ
AGRNS
AGUBO
AGWIK
AGYEJ
AHHHB
AHJVU
AHZHX
AIALX
AIEXJ
AIGII
AIIUN
AIKHN
AITUG
AKBMS
AKRWK
AKYEP
ALMA_UNASSIGNED_HOLDINGS
AMRAJ
ANKPU
AOUOD
APXCP
ARUGR
ASPBG
AVWKF
AXJTR
AZFZN
BJAXD
BKOJK
BLXMC
BNPGV
CS3
DU5
EBS
EFJIC
EJD
EO8
EO9
EP2
EP3
F0J
F5P
FDB
FEDTE
FGOYB
FIRID
FNPLU
FYGXN
G-2
G-Q
GBLVA
GBOLZ
HLZ
HMQ
HVGLF
HZ~
IHE
J1W
JJJVA
K-O
KOM
KZ1
LG9
LMP
M2V
M41
MHUIS
MO0
MOBAO
MVM
N9A
O-L
O9-
OAUVE
OZT
P-8
P-9
P2P
PC.
Q38
R2-
RIG
ROL
RPZ
SBC
SCC
SDF
SDG
SDP
SES
SEW
SNS
SPC
SPCBC
SSH
SSN
SST
SSV
SSW
SSZ
T5K
TAE
UAP
UNMZH
VOH
WUQ
XPP
ZMT
~G-
9DU
AAYXX
ACLOT
CITATION
EFKBS
EFLBG
~HD
CGR
CUY
CVF
ECM
EIF
NPM
7X8
ID FETCH-LOGICAL-c408t-5dbeceba3aed8efa1d9e8dada328ebea8628f954eddbd0590fc76cf9267cbb913
ISICitedReferencesCount 7
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001417182100001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 0893-6080
1879-2782
IngestDate Tue Sep 30 21:17:16 EDT 2025
Mon Jul 21 05:55:35 EDT 2025
Sat Nov 29 07:57:49 EST 2025
Tue Nov 18 22:40:46 EST 2025
Sat Jun 07 17:02:19 EDT 2025
IsDoiOpenAccess true
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Keywords Deep neural networks
Generative AI
Tabular data synthesis
Latent space interpolation
Wasserstein Autoencoder
Language English
License This is an open access article under the CC BY license.
Copyright © 2025 The Authors. Published by Elsevier Ltd.. All rights reserved.
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-c408t-5dbeceba3aed8efa1d9e8dada328ebea8628f954eddbd0590fc76cf9267cbb913
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ORCID 0000-0002-3691-8652
0000-0001-6203-6664
OpenAccessLink https://dx.doi.org/10.1016/j.neunet.2025.107208
PMID 39893805
PQID 3162846144
PQPubID 23479
ParticipantIDs proquest_miscellaneous_3162846144
pubmed_primary_39893805
crossref_citationtrail_10_1016_j_neunet_2025_107208
crossref_primary_10_1016_j_neunet_2025_107208
elsevier_sciencedirect_doi_10_1016_j_neunet_2025_107208
PublicationCentury 2000
PublicationDate May 2025
2025-05-00
2025-May
20250501
PublicationDateYYYYMMDD 2025-05-01
PublicationDate_xml – month: 05
  year: 2025
  text: May 2025
PublicationDecade 2020
PublicationPlace United States
PublicationPlace_xml – name: United States
PublicationTitle Neural networks
PublicationTitleAlternate Neural Netw
PublicationYear 2025
Publisher Elsevier Ltd
Publisher_xml – name: Elsevier Ltd
References (pp. 5902–5911).
Kim, J., Lee, C., & Park, N. (2023). STaSy: Score-based Tabular data Synthesis. In
Zhao, J., Kim, Y., Zhang, K., Rush, A., & LeCun, Y. (2018). Adversarially regularized autoencoders. In
(pp. 7176–7185).
Kingma, D. P., & Welling, M. (2014). Auto-encoding variational Bayes. In
Wang, Chukova, Simpson, Nguyen (b33) 2024; 166
Kim, J., Lee, C., Shin, Y., Park, S., Kim, M., Park, N., et al. (2022). SOS: Score-based oversampling for tabular data. In
Wang, Chukova, Nguyen (b31) 2023; 629
Borisov, Leemann, Seßler, Haug, Pawelczyk, Kasneci (b3) 2022
Woo, Reiter, Oganian, Karr (b36) 2009; 1
James, Harbron, Branson, Sundler (b12) 2021; 1
Yan, Yan, Wan, Zhang, Omberg, Guinney (b38) 2022; 13
Beckham, Honari, Verma, Lamb, Ghadiri, Hjelm (b2) 2019; 32
(pp. 97–112).
Raghunathan (b24) 2021; 8
Sun, Y., Cuesta-Infante, A., & Veeramachaneni, K. (2019). Learning vine copula models for synthetic data generation.
Naeem, M. F., Oh, S. J., Uh, Y., Choi, Y., & Yoo, J. (2020). Reliable fidelity and diversity metrics for generative models. In
Torfi, Fox, Reddy (b30) 2022; 586
Lampis, Lomurno, Matteucci (b19) 2023
Struski, Sadowski, Danel, Tabor, Podolak (b27) 2023
Zhao, Birke, Chen (b42) 2023
.
Wang, Chukova, Nguyen (b32) 2023; 148
Rabbani, Samad (b23) 2023
Borisov, V., Seßler, K., Leemann, T., Pawelczyk, M., & Kasneci, G. (2023). Language models are realistic tabular data generators. In
Wang, Simpson, Nguyen (b35) 2025; 691
Tolstikhin, I., Bousquet, O., Gelly, S., & Schoelkopf, B. (2018). A note on the evaluation of generative models. In
Zhao, Z., Birke, R., & Chen, L. Y. (2023a). FCT-GAN: Enhancing Global Correlation of Table Synthesis via Fourier Transform. In
(pp. 17564–17579).
Chong, M. J., & Forsyth, D. (2020). Effectively unbiased fid and inception score and where to find them. In
In
(pp. 4450–4454).
Shwartz-Ziv, Armon (b25) 2022; 81
ICLR 2024.
(pp. 1–18).
Xu, Skoularidou, Cuesta-Infante, Veeramachaneni (b37) 2019
Dablain, Krawczyk, Chawla (b8) 2022
Wang, Chukova, Sporle, Milne, Simpson, Nguyen (b34) 2024; 61
Lai, C.-H., Zou, D., & Lerman, G. (2023). Robust Variational Autoencoding with Wasserstein Penalty for Novelty Detection. In
(pp. 762–772).
(pp. 3538–3567).
(pp. 4610–4617).
(pp. 1–16).
(pp. 5049–5057).
Kolouri, S., Pope, P. E., Martin, C. E., & Rohde, G. K. (2019). Sliced Wasserstein auto-encoders. In
Platzer, Reutterer (b22) 2021
Gai, Zhang (b10) 2023
(pp. 6070–6079).
Yi, M., & Liu, S. (2023). Sliced Wasserstein variational inference. In
Zhao, Z., Kunar, A., Birke, R., & Chen, L. Y. (2021). CTAB-GAN: Effective table data synthesizing. In
ICLR 2023.
Zhang, H., Zhang, J., Srinivasan, B., Shen, Z., Qin, X., Faloutsos, C., et al. (2024). Mixed-Type Tabular Data Synthesis with Score-based Diffusion in Latent Space. In
Arbel, Korba, Salim, Gretton (b1) 2019; 32
Solatorio, Dupriez (b26) 2023
Kotelnikov, A., Baranchuk, D., Rubachev, I., & Babenko, A. (2023). TABDDPM: Modelling tabular data with diffusion models. In
Bousquet, Gelly, Tolstikhin, Simon-Gabriel, Schoelkopf (b5) 2017
Gupta, A., Bhatt, D., & Pandey, A. (2021). Transitioning from Real to Synthetic data: Quantifying the bias in model. In
Mukherjee, S., Asnani, H., Lin, E., & Kannan, S. (2019). ClusterGAN: Latent Space Clustering in Generative Adversarial Networks.
(pp. 1213–1228).
Zhao, Kunar, Birke, Chen (b45) 2022
Fonseca, Bacao (b9) 2023; 10
Chawla, Bowyer, Hall, Kegelmeyer (b6) 2002; 16
Dablain (10.1016/j.neunet.2025.107208_b8) 2022
Chawla (10.1016/j.neunet.2025.107208_b6) 2002; 16
Fonseca (10.1016/j.neunet.2025.107208_b9) 2023; 10
Solatorio (10.1016/j.neunet.2025.107208_b26) 2023
Wang (10.1016/j.neunet.2025.107208_b32) 2023; 148
Wang (10.1016/j.neunet.2025.107208_b35) 2025; 691
Woo (10.1016/j.neunet.2025.107208_b36) 2009; 1
Platzer (10.1016/j.neunet.2025.107208_b22) 2021
James (10.1016/j.neunet.2025.107208_b12) 2021; 1
Raghunathan (10.1016/j.neunet.2025.107208_b24) 2021; 8
Xu (10.1016/j.neunet.2025.107208_b37) 2019
10.1016/j.neunet.2025.107208_b44
Gai (10.1016/j.neunet.2025.107208_b10) 2023
10.1016/j.neunet.2025.107208_b21
10.1016/j.neunet.2025.107208_b43
10.1016/j.neunet.2025.107208_b20
10.1016/j.neunet.2025.107208_b41
Borisov (10.1016/j.neunet.2025.107208_b3) 2022
10.1016/j.neunet.2025.107208_b29
10.1016/j.neunet.2025.107208_b28
Wang (10.1016/j.neunet.2025.107208_b34) 2024; 61
Rabbani (10.1016/j.neunet.2025.107208_b23) 2023
Wang (10.1016/j.neunet.2025.107208_b31) 2023; 629
Zhao (10.1016/j.neunet.2025.107208_b42) 2023
Arbel (10.1016/j.neunet.2025.107208_b1) 2019; 32
Shwartz-Ziv (10.1016/j.neunet.2025.107208_b25) 2022; 81
Torfi (10.1016/j.neunet.2025.107208_b30) 2022; 586
Yan (10.1016/j.neunet.2025.107208_b38) 2022; 13
Lampis (10.1016/j.neunet.2025.107208_b19) 2023
Wang (10.1016/j.neunet.2025.107208_b33) 2024; 166
10.1016/j.neunet.2025.107208_b40
Bousquet (10.1016/j.neunet.2025.107208_b5) 2017
10.1016/j.neunet.2025.107208_b11
10.1016/j.neunet.2025.107208_b7
Zhao (10.1016/j.neunet.2025.107208_b45) 2022
10.1016/j.neunet.2025.107208_b15
Struski (10.1016/j.neunet.2025.107208_b27) 2023
10.1016/j.neunet.2025.107208_b14
10.1016/j.neunet.2025.107208_b4
10.1016/j.neunet.2025.107208_b13
10.1016/j.neunet.2025.107208_b18
10.1016/j.neunet.2025.107208_b17
10.1016/j.neunet.2025.107208_b39
10.1016/j.neunet.2025.107208_b16
Beckham (10.1016/j.neunet.2025.107208_b2) 2019; 32
References_xml – volume: 1
  start-page: 111
  year: 2009
  end-page: 124
  ident: b36
  article-title: Global measures of data utility for microdata masked for disclosure limitation
  publication-title: Journal of Privacy and Confidentiality
– year: 2023
  ident: b27
  article-title: Feature-based interpolation and geodesics in the latent spaces of generative models
  publication-title: IEEE Transactions on Neural Networks and Learning Systems
– year: 2023
  ident: b19
  article-title: Bridging the gap: Enhancing the utility of synthetic data via post-processing techniques
– reference: Zhang, H., Zhang, J., Srinivasan, B., Shen, Z., Qin, X., Faloutsos, C., et al. (2024). Mixed-Type Tabular Data Synthesis with Score-based Diffusion in Latent Space. In
– volume: 166
  year: 2024
  ident: b33
  article-title: Challenges and opportunities of generative models on tabular data
  publication-title: Applied Soft Computing
– volume: 629
  start-page: 313
  year: 2023
  end-page: 323
  ident: b31
  article-title: Ensemble k-nearest neighbors based on centroid displacement
  publication-title: Information Sciences
– reference: Kim, J., Lee, C., Shin, Y., Park, S., Kim, M., Park, N., et al. (2022). SOS: Score-based oversampling for tabular data. In
– reference: (pp. 1–16).
– reference: (pp. 17564–17579).
– reference: Sun, Y., Cuesta-Infante, A., & Veeramachaneni, K. (2019). Learning vine copula models for synthetic data generation.
– reference: Zhao, J., Kim, Y., Zhang, K., Rush, A., & LeCun, Y. (2018). Adversarially regularized autoencoders. In
– year: 2022
  ident: b3
  article-title: Deep neural networks and tabular data: A survey
  publication-title: IEEE Transactions on Neural Networks and Learning Systems
– reference: Gupta, A., Bhatt, D., & Pandey, A. (2021). Transitioning from Real to Synthetic data: Quantifying the bias in model. In
– reference: Naeem, M. F., Oh, S. J., Uh, Y., Choi, Y., & Yoo, J. (2020). Reliable fidelity and diversity metrics for generative models. In
– reference: Kingma, D. P., & Welling, M. (2014). Auto-encoding variational Bayes. In
– reference: Kim, J., Lee, C., & Park, N. (2023). STaSy: Score-based Tabular data Synthesis. In
– reference: Zhao, Z., Kunar, A., Birke, R., & Chen, L. Y. (2021). CTAB-GAN: Effective table data synthesizing. In
– year: 2023
  ident: b42
  article-title: Tabula: Harnessing language models for tabular data synthesis
– reference: (pp. 1–18).
– reference: (pp. 3538–3567).
– volume: 148
  year: 2023
  ident: b32
  article-title: Synthetic minority oversampling using edited displacement-based k-nearest neighbors
  publication-title: Applied Soft Computing
– volume: 32
  year: 2019
  ident: b1
  article-title: Maximum mean discrepancy gradient flow
  publication-title: Advances in Neural Information Processing Systems
– year: 2017
  ident: b5
  article-title: From optimal transport to generative modeling: the VEGAN cookbook
– reference: (pp. 762–772).
– reference: . ICLR 2024.
– year: 2022
  ident: b45
  article-title: CTAB-GAN+: Enhancing tabular data synthesis
– volume: 1
  start-page: 15
  year: 2021
  ident: b12
  article-title: Synthetic data use: exploring use cases to optimise data utility
  publication-title: Discover Artificial Intelligence
– reference: Yi, M., & Liu, S. (2023). Sliced Wasserstein variational inference. In
– reference: Kotelnikov, A., Baranchuk, D., Rubachev, I., & Babenko, A. (2023). TABDDPM: Modelling tabular data with diffusion models. In
– reference: (pp. 4450–4454).
– volume: 16
  start-page: 321
  year: 2002
  end-page: 357
  ident: b6
  article-title: SMOTE: synthetic minority over-sampling technique
  publication-title: Journal of Artificial Intelligence Research
– reference: . ICLR 2023.
– year: 2023
  ident: b10
  article-title: Tessellating the latent space for non-adversarial generative auto-encoders
  publication-title: IEEE Transactions on Pattern Analysis and Machine Intelligence
– volume: 32
  year: 2019
  ident: b2
  article-title: On adversarial mixup resynthesis
  publication-title: Advances in Neural Information Processing Systems
– reference: (pp. 5049–5057).
– volume: 13
  start-page: 7609
  year: 2022
  ident: b38
  article-title: A multifaceted benchmarking of synthetic electronic health record generation models
  publication-title: Nature Communications
– reference: (pp. 97–112).
– reference: (pp. 7176–7185).
– year: 2023
  ident: b26
  article-title: Realtabformer: Generating realistic relational and tabular data using transformers
– reference: Mukherjee, S., Asnani, H., Lin, E., & Kannan, S. (2019). ClusterGAN: Latent Space Clustering in Generative Adversarial Networks.
– year: 2021
  ident: b22
  article-title: Holdout-based fidelity and privacy assessment of mixed-type synthetic data
– reference: Chong, M. J., & Forsyth, D. (2020). Effectively unbiased fid and inception score and where to find them. In
– reference: Tolstikhin, I., Bousquet, O., Gelly, S., & Schoelkopf, B. (2018). A note on the evaluation of generative models. In
– reference: Zhao, Z., Birke, R., & Chen, L. Y. (2023a). FCT-GAN: Enhancing Global Correlation of Table Synthesis via Fourier Transform. In
– reference: Kolouri, S., Pope, P. E., Martin, C. E., & Rohde, G. K. (2019). Sliced Wasserstein auto-encoders. In
– volume: 81
  start-page: 84
  year: 2022
  end-page: 90
  ident: b25
  article-title: Tabular data: Deep learning is not all you need
  publication-title: Information Fusion
– reference: Borisov, V., Seßler, K., Leemann, T., Pawelczyk, M., & Kasneci, G. (2023). Language models are realistic tabular data generators. In
– volume: 8
  year: 2021
  ident: b24
  article-title: Synthetic data
  publication-title: Annual Review of Statistics and its Application
– reference: (pp. 6070–6079).
– year: 2023
  ident: b23
  article-title: Between-sample relationship in learning tabular data using graph and attention networks
– volume: 61
  year: 2024
  ident: b34
  article-title: Enhancing public research on citizen data: An empirical investigation of data synthesis using Statistics New Zealand’s Integrated Data Infrastructure
  publication-title: Information Processing & Management
– reference: (pp. 1213–1228).
– start-page: 7335
  year: 2019
  end-page: 7345
  ident: b37
  article-title: Modeling tabular data using conditional GAN
  publication-title: Advances in neural information processing systems
– reference: , In
– reference: (pp. 5902–5911).
– reference: .
– volume: 586
  start-page: 485
  year: 2022
  end-page: 500
  ident: b30
  article-title: Differentially private synthetic medical data generation using convolutional GANs
  publication-title: Information Sciences
– reference: (pp. 4610–4617).
– volume: 691
  year: 2025
  ident: b35
  article-title: Blending is all you need: Data-centric ensemble synthetic data
  publication-title: Information Sciences
– reference: Lai, C.-H., Zou, D., & Lerman, G. (2023). Robust Variational Autoencoding with Wasserstein Penalty for Novelty Detection. In
– start-page: 1
  year: 2022
  end-page: 15
  ident: b8
  article-title: DeepSMOTE: Fusing deep learning and SMOTE for imbalanced data
  publication-title: IEEE Transactions on Neural Networks and Learning Systems
– volume: 10
  start-page: 115
  year: 2023
  ident: b9
  article-title: Tabular and latent space synthetic data generation: a literature review
  publication-title: Journal of Big Data
– volume: 166
  year: 2024
  ident: 10.1016/j.neunet.2025.107208_b33
  article-title: Challenges and opportunities of generative models on tabular data
  publication-title: Applied Soft Computing
  doi: 10.1016/j.asoc.2024.112223
– year: 2023
  ident: 10.1016/j.neunet.2025.107208_b42
– volume: 10
  start-page: 115
  issue: 1
  year: 2023
  ident: 10.1016/j.neunet.2025.107208_b9
  article-title: Tabular and latent space synthetic data generation: a literature review
  publication-title: Journal of Big Data
  doi: 10.1186/s40537-023-00792-7
– ident: 10.1016/j.neunet.2025.107208_b14
  doi: 10.1145/3534678.3539454
– volume: 32
  year: 2019
  ident: 10.1016/j.neunet.2025.107208_b2
  article-title: On adversarial mixup resynthesis
  publication-title: Advances in Neural Information Processing Systems
– year: 2022
  ident: 10.1016/j.neunet.2025.107208_b3
  article-title: Deep neural networks and tabular data: A survey
  publication-title: IEEE Transactions on Neural Networks and Learning Systems
– year: 2023
  ident: 10.1016/j.neunet.2025.107208_b10
  article-title: Tessellating the latent space for non-adversarial generative auto-encoders
  publication-title: IEEE Transactions on Pattern Analysis and Machine Intelligence
– ident: 10.1016/j.neunet.2025.107208_b28
  doi: 10.1609/aaai.v33i01.33015049
– volume: 1
  start-page: 15
  issue: 1
  year: 2021
  ident: 10.1016/j.neunet.2025.107208_b12
  article-title: Synthetic data use: exploring use cases to optimise data utility
  publication-title: Discover Artificial Intelligence
  doi: 10.1007/s44163-021-00016-y
– year: 2023
  ident: 10.1016/j.neunet.2025.107208_b19
– volume: 16
  start-page: 321
  year: 2002
  ident: 10.1016/j.neunet.2025.107208_b6
  article-title: SMOTE: synthetic minority over-sampling technique
  publication-title: Journal of Artificial Intelligence Research
  doi: 10.1613/jair.953
– volume: 148
  year: 2023
  ident: 10.1016/j.neunet.2025.107208_b32
  article-title: Synthetic minority oversampling using edited displacement-based k-nearest neighbors
  publication-title: Applied Soft Computing
  doi: 10.1016/j.asoc.2023.110895
– volume: 586
  start-page: 485
  year: 2022
  ident: 10.1016/j.neunet.2025.107208_b30
  article-title: Differentially private synthetic medical data generation using convolutional GANs
  publication-title: Information Sciences
  doi: 10.1016/j.ins.2021.12.018
– ident: 10.1016/j.neunet.2025.107208_b17
– year: 2023
  ident: 10.1016/j.neunet.2025.107208_b26
– year: 2022
  ident: 10.1016/j.neunet.2025.107208_b45
– start-page: 1
  year: 2022
  ident: 10.1016/j.neunet.2025.107208_b8
  article-title: DeepSMOTE: Fusing deep learning and SMOTE for imbalanced data
  publication-title: IEEE Transactions on Neural Networks and Learning Systems
– ident: 10.1016/j.neunet.2025.107208_b21
– ident: 10.1016/j.neunet.2025.107208_b41
  doi: 10.1145/3583780.3615202
– ident: 10.1016/j.neunet.2025.107208_b15
– ident: 10.1016/j.neunet.2025.107208_b20
  doi: 10.1609/aaai.v33i01.33014610
– ident: 10.1016/j.neunet.2025.107208_b40
– year: 2021
  ident: 10.1016/j.neunet.2025.107208_b22
– ident: 10.1016/j.neunet.2025.107208_b44
– volume: 1
  start-page: 111
  issue: 1
  year: 2009
  ident: 10.1016/j.neunet.2025.107208_b36
  article-title: Global measures of data utility for microdata masked for disclosure limitation
  publication-title: Journal of Privacy and Confidentiality
  doi: 10.29012/jpc.v1i1.568
– ident: 10.1016/j.neunet.2025.107208_b13
– volume: 691
  year: 2025
  ident: 10.1016/j.neunet.2025.107208_b35
  article-title: Blending is all you need: Data-centric ensemble synthetic data
  publication-title: Information Sciences
  doi: 10.1016/j.ins.2024.121610
– ident: 10.1016/j.neunet.2025.107208_b11
– ident: 10.1016/j.neunet.2025.107208_b29
– volume: 8
  year: 2021
  ident: 10.1016/j.neunet.2025.107208_b24
  article-title: Synthetic data
  publication-title: Annual Review of Statistics and its Application
  doi: 10.1146/annurev-statistics-040720-031848
– start-page: 7335
  year: 2019
  ident: 10.1016/j.neunet.2025.107208_b37
  article-title: Modeling tabular data using conditional GAN
– year: 2023
  ident: 10.1016/j.neunet.2025.107208_b27
  article-title: Feature-based interpolation and geodesics in the latent spaces of generative models
  publication-title: IEEE Transactions on Neural Networks and Learning Systems
– volume: 81
  start-page: 84
  year: 2022
  ident: 10.1016/j.neunet.2025.107208_b25
  article-title: Tabular data: Deep learning is not all you need
  publication-title: Information Fusion
  doi: 10.1016/j.inffus.2021.11.011
– volume: 13
  start-page: 7609
  issue: 1
  year: 2022
  ident: 10.1016/j.neunet.2025.107208_b38
  article-title: A multifaceted benchmarking of synthetic electronic health record generation models
  publication-title: Nature Communications
  doi: 10.1038/s41467-022-35295-1
– volume: 629
  start-page: 313
  year: 2023
  ident: 10.1016/j.neunet.2025.107208_b31
  article-title: Ensemble k-nearest neighbors based on centroid displacement
  publication-title: Information Sciences
  doi: 10.1016/j.ins.2023.02.004
– ident: 10.1016/j.neunet.2025.107208_b4
– ident: 10.1016/j.neunet.2025.107208_b16
– year: 2023
  ident: 10.1016/j.neunet.2025.107208_b23
– ident: 10.1016/j.neunet.2025.107208_b39
– ident: 10.1016/j.neunet.2025.107208_b18
– ident: 10.1016/j.neunet.2025.107208_b43
– volume: 61
  issue: 1
  year: 2024
  ident: 10.1016/j.neunet.2025.107208_b34
  article-title: Enhancing public research on citizen data: An empirical investigation of data synthesis using Statistics New Zealand’s Integrated Data Infrastructure
  publication-title: Information Processing & Management
  doi: 10.1016/j.ipm.2023.103558
– ident: 10.1016/j.neunet.2025.107208_b7
  doi: 10.1109/CVPR42600.2020.00611
– volume: 32
  year: 2019
  ident: 10.1016/j.neunet.2025.107208_b1
  article-title: Maximum mean discrepancy gradient flow
  publication-title: Advances in Neural Information Processing Systems
– year: 2017
  ident: 10.1016/j.neunet.2025.107208_b5
SSID ssj0006843
Score 2.4973443
Snippet Tabular data generation is a complex task due to its distinctive characteristics and inherent complexities. While Variational Autoencoders have been adapted...
SourceID proquest
pubmed
crossref
elsevier
SourceType Aggregation Database
Index Database
Enrichment Source
Publisher
StartPage 107208
SubjectTerms Algorithms
Autoencoder
Deep Learning
Deep neural networks
Generative AI
Humans
Latent space interpolation
Neural Networks, Computer
Tabular data synthesis
Wasserstein Autoencoder
Title Deterministic Autoencoder using Wasserstein loss for tabular data generation
URI https://dx.doi.org/10.1016/j.neunet.2025.107208
https://www.ncbi.nlm.nih.gov/pubmed/39893805
https://www.proquest.com/docview/3162846144
Volume 185
WOSCitedRecordID wos001417182100001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVESC
  databaseName: Elsevier SD Freedom Collection Journals 2021
  customDbUrl:
  eissn: 1879-2782
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0006843
  issn: 0893-6080
  databaseCode: AIEXJ
  dateStart: 19950101
  isFulltext: true
  titleUrlDefault: https://www.sciencedirect.com
  providerName: Elsevier
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV1LbxMxELZoyoEL70cKVEZCXCJH2VdsHwOkAhQFDindm-W1vdCo2oQkW5V_z_ixm1SlajlwWa2s9drrzx6PZ2fmQ-htGcWSK1MQbXhGUtgjSKF0RFiRSqnoQBlHyfJ9QqdTluf8W2CzXzs6AVpV7OKCL_8r1FAGYNvQ2X-Au30pFMA9gA5XgB2utwL-Y3BwcRmYe6N6s7C5Km3KiNrZBU6ki7C0LJe9M9givaOhLJw_qnUYtazKZrVFbN4keHIZOirvN96q4ifB4GwjZXp5a1v-Uf_2Au39afUzBJEF40KcbV35gjxklJOYen6gvvlLWStEsx0xCGfK2KVruCqhvbFg3q9MDR3u20b728cvJ8SefhVHx5OJmI3z2bvlL2K5wuw_9UCcsof2Y5px1kH7o8_j_Eu7Aw-ZD6xoOtqETDq_vqsNX6eSXHfkcKrH7CG6H84MeOSxfoTumOoxetDwceAgnp-gySXo8Q702EGPd6DHFnoM0OMAPbbQ4y30T9Hx0Xj24RMJbBlEpQO2IZmG5WgKmUijmSllpLlhWmqZxAxWqoSjKyt5lhqtC21DjktFh6rk8ZCqouBR8gx1qkVlXiCcREPDBiyFxRqnTBdMGZ2qjCpegPoroy5KmgETKqSSt4wmZ6LxGZwLP8zCDrPww9xFpK219KlUbnieNliIoA56NU_AXLqh5psGOgHS0v4Ck5VZ1GsBXwb6mDWCdNFzj2nbl4SD7s4G2cEtar9E97bL5RXqbFa1eY3uqvPN6Xp1iPZozg7DrPwDj2CZOg
linkProvider Elsevier
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Deterministic+Autoencoder+using+Wasserstein+loss+for+tabular+data+generation&rft.jtitle=Neural+networks&rft.au=Wang%2C+Alex+X&rft.au=Nguyen%2C+Binh+P&rft.date=2025-05-01&rft.issn=1879-2782&rft.eissn=1879-2782&rft.volume=185&rft.spage=107208&rft_id=info:doi/10.1016%2Fj.neunet.2025.107208&rft.externalDBID=NO_FULL_TEXT
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=0893-6080&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=0893-6080&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=0893-6080&client=summon