EBLearn

Open-source energy-based learning in C++

Pierre Sermanet, Koray Kavukcuoglu, Yann LeCun

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

Abstract

Energy-based learning (EBL) is a general framework to describe supervised and unsupervised training methods for probabilistic and non-probabilistic factor graphs. An energy-based model associates a scalar energy to configurations of inputs, outputs, and latent variables. Learning machines can be constructed by assembling modules and loss functions. Gradient-based learning procedures are easily implemented through semi-automatic differentiation of complex models constructed by assembling predefined modules. We introduce an open-source and cross-platform C++ library called EBLearn1 to enable the construction of energy-based learning models. EBLearn is composed of two major components, libidx: an efficient and flexible multi-dimensional tensor library, and libeblearn: an object-oriented library of trainable modules and learning algorithms. The latter has facilities for such models as convolutional networks, as well as for image processing. It also provides graphical display functions.

Original languageEnglish (US)
Title of host publicationICTAI 2009 - 21st IEEE International Conference on Tools with Artificial Intelligence
Pages693-697
Number of pages5
DOIs
StatePublished - 2009
Event21st IEEE International Conference on Tools with Artificial Intelligence, ICTAI 2009 - Newark, NJ, United States
Duration: Nov 2 2009Nov 5 2009

Other

Other21st IEEE International Conference on Tools with Artificial Intelligence, ICTAI 2009
CountryUnited States
CityNewark, NJ
Period11/2/0911/5/09

Fingerprint

Learning algorithms
Tensors
Learning systems
Image processing
Display devices

ASJC Scopus subject areas

  • Software
  • Artificial Intelligence
  • Computer Science Applications

Cite this

Sermanet, P., Kavukcuoglu, K., & LeCun, Y. (2009). EBLearn: Open-source energy-based learning in C++. In ICTAI 2009 - 21st IEEE International Conference on Tools with Artificial Intelligence (pp. 693-697). [5366626] https://doi.org/10.1109/ICTAI.2009.28

EBLearn : Open-source energy-based learning in C++. / Sermanet, Pierre; Kavukcuoglu, Koray; LeCun, Yann.

ICTAI 2009 - 21st IEEE International Conference on Tools with Artificial Intelligence. 2009. p. 693-697 5366626.

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

Sermanet, P, Kavukcuoglu, K & LeCun, Y 2009, EBLearn: Open-source energy-based learning in C++. in ICTAI 2009 - 21st IEEE International Conference on Tools with Artificial Intelligence., 5366626, pp. 693-697, 21st IEEE International Conference on Tools with Artificial Intelligence, ICTAI 2009, Newark, NJ, United States, 11/2/09. https://doi.org/10.1109/ICTAI.2009.28
Sermanet P, Kavukcuoglu K, LeCun Y. EBLearn: Open-source energy-based learning in C++. In ICTAI 2009 - 21st IEEE International Conference on Tools with Artificial Intelligence. 2009. p. 693-697. 5366626 https://doi.org/10.1109/ICTAI.2009.28
Sermanet, Pierre ; Kavukcuoglu, Koray ; LeCun, Yann. / EBLearn : Open-source energy-based learning in C++. ICTAI 2009 - 21st IEEE International Conference on Tools with Artificial Intelligence. 2009. pp. 693-697
@inproceedings{01d188a656f74a8a82516f132d7eb5af,
title = "EBLearn: Open-source energy-based learning in C++",
abstract = "Energy-based learning (EBL) is a general framework to describe supervised and unsupervised training methods for probabilistic and non-probabilistic factor graphs. An energy-based model associates a scalar energy to configurations of inputs, outputs, and latent variables. Learning machines can be constructed by assembling modules and loss functions. Gradient-based learning procedures are easily implemented through semi-automatic differentiation of complex models constructed by assembling predefined modules. We introduce an open-source and cross-platform C++ library called EBLearn1 to enable the construction of energy-based learning models. EBLearn is composed of two major components, libidx: an efficient and flexible multi-dimensional tensor library, and libeblearn: an object-oriented library of trainable modules and learning algorithms. The latter has facilities for such models as convolutional networks, as well as for image processing. It also provides graphical display functions.",
author = "Pierre Sermanet and Koray Kavukcuoglu and Yann LeCun",
year = "2009",
doi = "10.1109/ICTAI.2009.28",
language = "English (US)",
isbn = "9781424456192",
pages = "693--697",
booktitle = "ICTAI 2009 - 21st IEEE International Conference on Tools with Artificial Intelligence",

}

TY - GEN

T1 - EBLearn

T2 - Open-source energy-based learning in C++

AU - Sermanet, Pierre

AU - Kavukcuoglu, Koray

AU - LeCun, Yann

PY - 2009

Y1 - 2009

N2 - Energy-based learning (EBL) is a general framework to describe supervised and unsupervised training methods for probabilistic and non-probabilistic factor graphs. An energy-based model associates a scalar energy to configurations of inputs, outputs, and latent variables. Learning machines can be constructed by assembling modules and loss functions. Gradient-based learning procedures are easily implemented through semi-automatic differentiation of complex models constructed by assembling predefined modules. We introduce an open-source and cross-platform C++ library called EBLearn1 to enable the construction of energy-based learning models. EBLearn is composed of two major components, libidx: an efficient and flexible multi-dimensional tensor library, and libeblearn: an object-oriented library of trainable modules and learning algorithms. The latter has facilities for such models as convolutional networks, as well as for image processing. It also provides graphical display functions.

AB - Energy-based learning (EBL) is a general framework to describe supervised and unsupervised training methods for probabilistic and non-probabilistic factor graphs. An energy-based model associates a scalar energy to configurations of inputs, outputs, and latent variables. Learning machines can be constructed by assembling modules and loss functions. Gradient-based learning procedures are easily implemented through semi-automatic differentiation of complex models constructed by assembling predefined modules. We introduce an open-source and cross-platform C++ library called EBLearn1 to enable the construction of energy-based learning models. EBLearn is composed of two major components, libidx: an efficient and flexible multi-dimensional tensor library, and libeblearn: an object-oriented library of trainable modules and learning algorithms. The latter has facilities for such models as convolutional networks, as well as for image processing. It also provides graphical display functions.

UR - http://www.scopus.com/inward/record.url?scp=77949515497&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77949515497&partnerID=8YFLogxK

U2 - 10.1109/ICTAI.2009.28

DO - 10.1109/ICTAI.2009.28

M3 - Conference contribution

SN - 9781424456192

SP - 693

EP - 697

BT - ICTAI 2009 - 21st IEEE International Conference on Tools with Artificial Intelligence

ER -