Learning invariant features by harnessing the aperture problem

Roland Memisevic; Georgios Exarchakis

Learning invariant features by harnessing the aperture problem

Roland Memisevic, Georgios Exarchakis

Department of Computer Science

Research output: Contribution to conference › Paper › peer-review

3 Citations (SciVal)

Abstract

The energy model is a simple, biologically inspired approach to extracting relationships between images in tasks like stereopsis and motion analysis. We discuss how adding an extra pooling layer to the energy model makes it possible to learn encodings of transformations that are mostly invariant with respect to image content, and to learn encodings of images that are mostly invariant with respect to the observed transformations. We show how this makes it possible to learn 3D pose-invariant features of objects by watching videos of the objects. We test our approach on a dataset of videos derived from the NORB dataset.

Original language	English
Pages	1137-1145
Number of pages	9
Publication status	Published - 21 Jun 2013
Event	30th International Conference on Machine Learning, ICML 2013 - Atlanta, GA, USA United States Duration: 16 Jun 2013 → 21 Jun 2013

Conference

Conference	30th International Conference on Machine Learning, ICML 2013
Country/Territory	USA United States
City	Atlanta, GA
Period	16/06/13 → 21/06/13

ASJC Scopus subject areas

Human-Computer Interaction
Sociology and Political Science

Access to Document

https://proceedings.mlr.press/v28/memisevic13.htmlLicence: Unspecified

Cite this

@conference{2115eda35aa242798bbf11269457b921,

title = "Learning invariant features by harnessing the aperture problem",

abstract = "The energy model is a simple, biologically inspired approach to extracting relationships between images in tasks like stereopsis and motion analysis. We discuss how adding an extra pooling layer to the energy model makes it possible to learn encodings of transformations that are mostly invariant with respect to image content, and to learn encodings of images that are mostly invariant with respect to the observed transformations. We show how this makes it possible to learn 3D pose-invariant features of objects by watching videos of the objects. We test our approach on a dataset of videos derived from the NORB dataset.",

author = "Roland Memisevic and Georgios Exarchakis",

year = "2013",

month = jun,

day = "21",

language = "English",

pages = "1137--1145",

note = "30th International Conference on Machine Learning, ICML 2013 ; Conference date: 16-06-2013 Through 21-06-2013",

}

TY - CONF

T1 - Learning invariant features by harnessing the aperture problem

AU - Memisevic, Roland

AU - Exarchakis, Georgios

PY - 2013/6/21

Y1 - 2013/6/21

N2 - The energy model is a simple, biologically inspired approach to extracting relationships between images in tasks like stereopsis and motion analysis. We discuss how adding an extra pooling layer to the energy model makes it possible to learn encodings of transformations that are mostly invariant with respect to image content, and to learn encodings of images that are mostly invariant with respect to the observed transformations. We show how this makes it possible to learn 3D pose-invariant features of objects by watching videos of the objects. We test our approach on a dataset of videos derived from the NORB dataset.

AB - The energy model is a simple, biologically inspired approach to extracting relationships between images in tasks like stereopsis and motion analysis. We discuss how adding an extra pooling layer to the energy model makes it possible to learn encodings of transformations that are mostly invariant with respect to image content, and to learn encodings of images that are mostly invariant with respect to the observed transformations. We show how this makes it possible to learn 3D pose-invariant features of objects by watching videos of the objects. We test our approach on a dataset of videos derived from the NORB dataset.

UR - http://www.scopus.com/inward/record.url?scp=84897514029&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:84897514029

SP - 1137

EP - 1145

T2 - 30th International Conference on Machine Learning, ICML 2013

Y2 - 16 June 2013 through 21 June 2013

ER -

Learning invariant features by harnessing the aperture problem

Abstract

Conference

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this