The HPU

James Davis, Joan Arderiu, Henry Lin, Zeb Nevins, Sebastian Schuon, Orazio Gallo, Ming Hsuan Yang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

9 Citations (Scopus)

Abstract

Computer-mediated, human micro-labor markets have so far been treated as novelty services good for cheaply labeling training data and easy user studies. This paper's primary contribution is conceptual, the claim that these markets can be characterized as Human co-Processing Units (HPU), and represent a first class computational platform. In the same way that Graphics Processing Units (GPU) represent a change in architecture from CPU based computation, HPU based computation is different, and deserves careful characterization and study. We demonstrate the value of this claim by showing that simplistic HPU computation can be more accurate than complex CPU based algorithms on some important computer vision tasks. We also argue that HPU computation can be cheaper than state-of-the-art CPU based computation. Finally we give some examples of characterizing the HPU as an architectural platform.

Original languageEnglish
Title of host publication2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010
Pages9-16
Number of pages8
DOIs
Publication statusPublished - 2010 Sep 17
Event2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010 - San Francisco, CA, United States
Duration: 2010 Jun 132010 Jun 18

Publication series

Name2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010

Conference

Conference2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010
CountryUnited States
CitySan Francisco, CA
Period10/6/1310/6/18

Fingerprint

Program processors
Processing
Labeling
Computer vision
Personnel
Graphics processing unit

All Science Journal Classification (ASJC) codes

  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering

Cite this

Davis, J., Arderiu, J., Lin, H., Nevins, Z., Schuon, S., Gallo, O., & Yang, M. H. (2010). The HPU. In 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010 (pp. 9-16). [5543192] (2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010). https://doi.org/10.1109/CVPRW.2010.5543192
Davis, James ; Arderiu, Joan ; Lin, Henry ; Nevins, Zeb ; Schuon, Sebastian ; Gallo, Orazio ; Yang, Ming Hsuan. / The HPU. 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010. 2010. pp. 9-16 (2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010).
@inproceedings{950b662e9e1449479cecaf9dff303ba8,
title = "The HPU",
abstract = "Computer-mediated, human micro-labor markets have so far been treated as novelty services good for cheaply labeling training data and easy user studies. This paper's primary contribution is conceptual, the claim that these markets can be characterized as Human co-Processing Units (HPU), and represent a first class computational platform. In the same way that Graphics Processing Units (GPU) represent a change in architecture from CPU based computation, HPU based computation is different, and deserves careful characterization and study. We demonstrate the value of this claim by showing that simplistic HPU computation can be more accurate than complex CPU based algorithms on some important computer vision tasks. We also argue that HPU computation can be cheaper than state-of-the-art CPU based computation. Finally we give some examples of characterizing the HPU as an architectural platform.",
author = "James Davis and Joan Arderiu and Henry Lin and Zeb Nevins and Sebastian Schuon and Orazio Gallo and Yang, {Ming Hsuan}",
year = "2010",
month = "9",
day = "17",
doi = "10.1109/CVPRW.2010.5543192",
language = "English",
isbn = "9781424470297",
series = "2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010",
pages = "9--16",
booktitle = "2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010",

}

Davis, J, Arderiu, J, Lin, H, Nevins, Z, Schuon, S, Gallo, O & Yang, MH 2010, The HPU. in 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010., 5543192, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010, pp. 9-16, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010, San Francisco, CA, United States, 10/6/13. https://doi.org/10.1109/CVPRW.2010.5543192

The HPU. / Davis, James; Arderiu, Joan; Lin, Henry; Nevins, Zeb; Schuon, Sebastian; Gallo, Orazio; Yang, Ming Hsuan.

2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010. 2010. p. 9-16 5543192 (2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - The HPU

AU - Davis, James

AU - Arderiu, Joan

AU - Lin, Henry

AU - Nevins, Zeb

AU - Schuon, Sebastian

AU - Gallo, Orazio

AU - Yang, Ming Hsuan

PY - 2010/9/17

Y1 - 2010/9/17

N2 - Computer-mediated, human micro-labor markets have so far been treated as novelty services good for cheaply labeling training data and easy user studies. This paper's primary contribution is conceptual, the claim that these markets can be characterized as Human co-Processing Units (HPU), and represent a first class computational platform. In the same way that Graphics Processing Units (GPU) represent a change in architecture from CPU based computation, HPU based computation is different, and deserves careful characterization and study. We demonstrate the value of this claim by showing that simplistic HPU computation can be more accurate than complex CPU based algorithms on some important computer vision tasks. We also argue that HPU computation can be cheaper than state-of-the-art CPU based computation. Finally we give some examples of characterizing the HPU as an architectural platform.

AB - Computer-mediated, human micro-labor markets have so far been treated as novelty services good for cheaply labeling training data and easy user studies. This paper's primary contribution is conceptual, the claim that these markets can be characterized as Human co-Processing Units (HPU), and represent a first class computational platform. In the same way that Graphics Processing Units (GPU) represent a change in architecture from CPU based computation, HPU based computation is different, and deserves careful characterization and study. We demonstrate the value of this claim by showing that simplistic HPU computation can be more accurate than complex CPU based algorithms on some important computer vision tasks. We also argue that HPU computation can be cheaper than state-of-the-art CPU based computation. Finally we give some examples of characterizing the HPU as an architectural platform.

UR - http://www.scopus.com/inward/record.url?scp=77956545410&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77956545410&partnerID=8YFLogxK

U2 - 10.1109/CVPRW.2010.5543192

DO - 10.1109/CVPRW.2010.5543192

M3 - Conference contribution

AN - SCOPUS:77956545410

SN - 9781424470297

T3 - 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010

SP - 9

EP - 16

BT - 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010

ER -

Davis J, Arderiu J, Lin H, Nevins Z, Schuon S, Gallo O et al. The HPU. In 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010. 2010. p. 9-16. 5543192. (2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Workshops, CVPRW 2010). https://doi.org/10.1109/CVPRW.2010.5543192