Sylvester normalizing flows for variational inference

Rianne Van Den Berg; Leonard Hasenclever; Jakub M. Tomczak; Max Welling

Sylvester normalizing flows for variational inference

Rianne Van Den Berg, Leonard Hasenclever, Jakub M. Tomczak, Max Welling

Research output: Chapter in Book / Report / Conference proceeding › Conference contribution › Academic › peer-review

Abstract

Variational inference relies on flexible approximate posterior distributions. Normalizing flows provide a general recipe to construct flexible variational posteriors. We introduce Sylvester normalizing flows, which can be seen as a generalization of planar flows. Sylvester normalizing flows remove the well-known single-unit bottleneck from planar flows, making a single transformation much more flexible. We compare the performance of Sylvester normalizing flows against planar flows and inverse autoregressive flows and demonstrate that they compare favorably on several datasets.

Original language	English
Title of host publication	34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018
Editors	Amir Globerson, Amir Globerson, Ricardo Silva
Publisher	Association For Uncertainty in Artificial Intelligence (AUAI)
Pages	393-402
Number of pages	10
ISBN (Electronic)	9781510871601
Publication status	Published - 1 Jan 2018
Externally published	Yes
Event	34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018 - Monterey, United States Duration: 6 Aug 2018 → 10 Aug 2018

Publication series

Name	34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018
Volume	1

Conference

Conference	34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018
Country/Territory	United States
City	Monterey
Period	6/08/18 → 10/08/18

Funding

We would like to thank Christos Louizos for useful discussions and helping with the implementation of inverse autoregressive flows. LH is funded by the UK EPSRC OxWaSP CDT through grant EP/L016710/1. JMT is funded by the European Commission within the MSC-IF (Grant No. 702666). RvdB is funded by SAP SE.

Funders	Funder number
MSC-IF
Engineering and Physical Sciences Research Council	EP/L016710/1
European Commission	702666

Persistent URL (handle)

Persistent link

Cite this

Van Den Berg, R., Hasenclever, L., Tomczak, J. M., & Welling, M. (2018). Sylvester normalizing flows for variational inference. In A. Globerson, A. Globerson, & R. Silva (Eds.), 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018 (pp. 393-402). (34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018; Vol. 1). Association For Uncertainty in Artificial Intelligence (AUAI).

Van Den Berg, Rianne ; Hasenclever, Leonard ; Tomczak, Jakub M. et al. / Sylvester normalizing flows for variational inference. 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018. editor / Amir Globerson ; Amir Globerson ; Ricardo Silva. Association For Uncertainty in Artificial Intelligence (AUAI), 2018. pp. 393-402 (34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018).

@inproceedings{74041561b21744fbb26beea08085dea0,

title = "Sylvester normalizing flows for variational inference",

abstract = "Variational inference relies on flexible approximate posterior distributions. Normalizing flows provide a general recipe to construct flexible variational posteriors. We introduce Sylvester normalizing flows, which can be seen as a generalization of planar flows. Sylvester normalizing flows remove the well-known single-unit bottleneck from planar flows, making a single transformation much more flexible. We compare the performance of Sylvester normalizing flows against planar flows and inverse autoregressive flows and demonstrate that they compare favorably on several datasets.",

author = "{Van Den Berg}, Rianne and Leonard Hasenclever and Tomczak, {Jakub M.} and Max Welling",

year = "2018",

month = jan,

day = "1",

language = "English",

series = "34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018",

publisher = "Association For Uncertainty in Artificial Intelligence (AUAI)",

pages = "393--402",

editor = "Amir Globerson and Amir Globerson and Ricardo Silva",

booktitle = "34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018",

note = "34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018 ; Conference date: 06-08-2018 Through 10-08-2018",

}

Van Den Berg, R, Hasenclever, L, Tomczak, JM & Welling, M 2018, Sylvester normalizing flows for variational inference. in A Globerson, A Globerson & R Silva (eds), 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018. 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018, vol. 1, Association For Uncertainty in Artificial Intelligence (AUAI), pp. 393-402, 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018, Monterey, United States, 6/08/18.

Sylvester normalizing flows for variational inference. / Van Den Berg, Rianne; Hasenclever, Leonard; Tomczak, Jakub M. et al.
34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018. ed. / Amir Globerson; Amir Globerson; Ricardo Silva. Association For Uncertainty in Artificial Intelligence (AUAI), 2018. p. 393-402 (34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018; Vol. 1).

Research output: Chapter in Book / Report / Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Sylvester normalizing flows for variational inference

AU - Van Den Berg, Rianne

AU - Hasenclever, Leonard

AU - Tomczak, Jakub M.

AU - Welling, Max

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Variational inference relies on flexible approximate posterior distributions. Normalizing flows provide a general recipe to construct flexible variational posteriors. We introduce Sylvester normalizing flows, which can be seen as a generalization of planar flows. Sylvester normalizing flows remove the well-known single-unit bottleneck from planar flows, making a single transformation much more flexible. We compare the performance of Sylvester normalizing flows against planar flows and inverse autoregressive flows and demonstrate that they compare favorably on several datasets.

AB - Variational inference relies on flexible approximate posterior distributions. Normalizing flows provide a general recipe to construct flexible variational posteriors. We introduce Sylvester normalizing flows, which can be seen as a generalization of planar flows. Sylvester normalizing flows remove the well-known single-unit bottleneck from planar flows, making a single transformation much more flexible. We compare the performance of Sylvester normalizing flows against planar flows and inverse autoregressive flows and demonstrate that they compare favorably on several datasets.

UR - http://www.scopus.com/inward/record.url?scp=85059413595&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85059413595&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85059413595

T3 - 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018

SP - 393

EP - 402

BT - 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018

A2 - Globerson, Amir

A2 - Silva, Ricardo

PB - Association For Uncertainty in Artificial Intelligence (AUAI)

T2 - 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018

Y2 - 6 August 2018 through 10 August 2018

ER -

Van Den Berg R, Hasenclever L, Tomczak JM, Welling M. Sylvester normalizing flows for variational inference. In Globerson A, Globerson A, Silva R, editors, 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018. Association For Uncertainty in Artificial Intelligence (AUAI). 2018. p. 393-402. (34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018).

Sylvester normalizing flows for variational inference

Abstract

Publication series

Conference

Funding

Persistent URL (handle)

Other files and links

Fingerprint

Cite this