A Self-Organizing Multiple-View Representations of 3D Objects.

Daphna Weinshall; Shimon Edelman; Heinrich H. Bülthoff

A Self-Organizing Multiple-View Representations of 3D Objects.

Daphna Weinshall, Shimon Edelman, Heinrich H. Bülthoff

The Rachel and Selim Benin School of Engineering and Computer Science

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

Abstract

We demonstrate the ability of a two-layer network of thresholded summation units to support representation of 3D objects in which several distinct 2D views are stored for ea.ch object. Using unsupervised Hebbian relaxation, the network learned to recognize ten objects from different viewpoints. The training process led to the emergence of compact representations of the specific input views. When tested on novel views of the same objects, the network exhibited a substantial generalization capability. In simulated psychophysical experiments, the network's behavior was qualitatively similar to that of human subjects.

Original language	English
Title of host publication	NIPS 1989
Editors	David S. Touretzky
Publisher	Morgan Kaufmann Publishers, Inc.
Pages	274-281
Number of pages	8
ISBN (Print)	9781558601000
State	Published - 1989
Event	3rd IEEE Conference on Neural Information Processing Systems, NIPS 1989 - Denver, United States Duration: 27 Nov 1989 → 30 Nov 1989 Conference number: 3

Publication series

Name	Advances in neural information processing systems
Publisher	Morgan Kaufmann Publishers
Volume	2
ISSN (Print)	1049-5258

Conference

Conference	3rd IEEE Conference on Neural Information Processing Systems, NIPS 1989
Abbreviated title	NIPS 1989
Country/Territory	United States
City	Denver
Period	27/11/89 → 30/11/89

Access to Document

Cite this

@inproceedings{0a25c7b7afcb42669ae10db9c1222ca0,

title = "A Self-Organizing Multiple-View Representations of 3D Objects.",

abstract = "We demonstrate the ability of a two-layer network of thresholded summation units to support representation of 3D objects in which several distinct 2D views are stored for ea.ch object. Using unsupervised Hebbian relaxation, the network learned to recognize ten objects from different viewpoints. The training process led to the emergence of compact representations of the specific input views. When tested on novel views of the same objects, the network exhibited a substantial generalization capability. In simulated psychophysical experiments, the network's behavior was qualitatively similar to that of human subjects.",

author = "Daphna Weinshall and Shimon Edelman and B{\"u}lthoff, {Heinrich H.}",

year = "1989",

language = "אנגלית",

isbn = "9781558601000",

series = "Advances in neural information processing systems",

publisher = "Morgan Kaufmann Publishers, Inc.",

pages = "274--281",

editor = "Touretzky, {David S.}",

booktitle = "NIPS 1989",

address = "ארצות הברית",

note = "3rd IEEE Conference on Neural Information Processing Systems, NIPS 1989 ; Conference date: 27-11-1989 Through 30-11-1989",

}

Weinshall, D, Edelman, S & Bülthoff, HH 1989, A Self-Organizing Multiple-View Representations of 3D Objects. in DS Touretzky (ed.), NIPS 1989. Advances in neural information processing systems, vol. 2, Morgan Kaufmann Publishers, Inc., pp. 274-281, 3rd IEEE Conference on Neural Information Processing Systems, NIPS 1989, Denver, Colorado, United States, 27/11/89. <https://dblp.org/db/conf/nips/nips1989.html#WeinshallEB89>

A Self-Organizing Multiple-View Representations of 3D Objects. / Weinshall, Daphna; Edelman, Shimon; Bülthoff, Heinrich H.
NIPS 1989. ed. / David S. Touretzky. Morgan Kaufmann Publishers, Inc., 1989. p. 274-281 (Advances in neural information processing systems; Vol. 2).

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

TY - GEN

T1 - A Self-Organizing Multiple-View Representations of 3D Objects.

AU - Weinshall, Daphna

AU - Edelman, Shimon

AU - Bülthoff, Heinrich H.

N1 - Conference code: 3

PY - 1989

Y1 - 1989

N2 - We demonstrate the ability of a two-layer network of thresholded summation units to support representation of 3D objects in which several distinct 2D views are stored for ea.ch object. Using unsupervised Hebbian relaxation, the network learned to recognize ten objects from different viewpoints. The training process led to the emergence of compact representations of the specific input views. When tested on novel views of the same objects, the network exhibited a substantial generalization capability. In simulated psychophysical experiments, the network's behavior was qualitatively similar to that of human subjects.

AB - We demonstrate the ability of a two-layer network of thresholded summation units to support representation of 3D objects in which several distinct 2D views are stored for ea.ch object. Using unsupervised Hebbian relaxation, the network learned to recognize ten objects from different viewpoints. The training process led to the emergence of compact representations of the specific input views. When tested on novel views of the same objects, the network exhibited a substantial generalization capability. In simulated psychophysical experiments, the network's behavior was qualitatively similar to that of human subjects.

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

SN - 9781558601000

T3 - Advances in neural information processing systems

SP - 274

EP - 281

BT - NIPS 1989

A2 - Touretzky, David S.

PB - Morgan Kaufmann Publishers, Inc.

T2 - 3rd IEEE Conference on Neural Information Processing Systems, NIPS 1989

Y2 - 27 November 1989 through 30 November 1989

ER -

A Self-Organizing Multiple-View Representations of 3D Objects.

Abstract

Publication series

Conference

Access to Document

Fingerprint

Cite this