PmWiki.SSL History
Hide minor edits - Show changes to output
Changed line 20 from:
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips2.ppt | PPT ]] |
to:
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips3.ppt | PPT ]] |
Changed line 20 from:
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips.ppt | PPT ]] |
to:
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips2.ppt | PPT ]] |
Changed line 25 from:
To see this algorithm running live on 80 million images, please visit the [[http://people.csail.mit.edu/torralba/tinyimages|Tiny Images]] webpage and try labeling a few examples.
to:
'''Live demo:''' To see this algorithm running live on 80 million images, please visit the [[http://people.csail.mit.edu/torralba/tinyimages|Tiny Images]] webpage and try labeling a few examples.
Added lines 22-25:
----
To see this algorithm running live on 80 million images, please visit the [[http://people.csail.mit.edu/torralba/tinyimages|Tiny Images]] webpage and try labeling a few examples.
Changed line 5 from:
With the advent of the Internet it is now possible to collect hundreds of millions
to:
'''Abstract:''' With the advent of the Internet it is now possible to collect hundreds of millions
Changed lines 19-20 from:
[[ http://cs.nyu.edu/~fergus/papers/fwt_ssl.pdf | PDF ]] \\
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips.ppt | PPT ]] \\
to:
[[ http://cs.nyu.edu/~fergus/papers/fwt_ssl.pdf | PDF ]] |
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips.ppt | PPT ]] |
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips.ppt | PPT ]] |
Changed lines 3-4 from:
[[http://cs.nyu.edu/~fergus | Rob Fergus]], [[http://www.cs.huji.ac.il/~yweiss/ | Yair Weiss]], [[http://web.mit.edu/torralba/www/ | Antonio Torralba]]
to:
[[http://cs.nyu.edu/~fergus | Rob Fergus]], [[http://www.cs.huji.ac.il/~yweiss/ | Yair Weiss]], [[http://web.mit.edu/torralba/www/ | Antonio Torralba]]
Changed lines 14-15 from:
for semi-supervised learning that are linear in the number of images. Specifically,
we use the convergence of the eigenvectors of the normalized graph Laplacian to
we use the convergence of the eigenvectors of the normalized graph Laplacian to
to:
for semi-supervised learning that are linear in the number of images. Specifically, we use the convergence of the eigenvectors of the normalized graph Laplacian to
Changed lines 19-20 from:
[[ http://cs.nyu.edu/~fergus/papers/fwt_ssl.pdf | PDF ]]
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips.ppt | PPT ]]
to:
[[ http://cs.nyu.edu/~fergus/papers/fwt_ssl.pdf | PDF ]] \\
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips.ppt | PPT ]] \\
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips.ppt | PPT ]] \\
Changed lines 3-22 from:
[[http://cs.nyu.edu/~fergus | Rob Fergus]], [[http://www.cs.huji.ac.il/~yweiss/ | Yair Weiss]], [[http://web.mit.edu/torralba/www/ | Antonio Torralba]]
to:
[[http://cs.nyu.edu/~fergus | Rob Fergus]], [[http://www.cs.huji.ac.il/~yweiss/ | Yair Weiss]], [[http://web.mit.edu/torralba/www/ | Antonio Torralba]]
With the advent of the Internet it is now possible to collect hundreds of millions
of images. These images come with varying degrees of label information. “Clean
labels” can be manually obtained on a small fraction, “noisy labels” may be extracted
automatically from surrounding text, while for most images there are no
labels at all. Semi-supervised learning is a principled framework for combining
these different label sources. However, it scales polynomially with the number
of images, making it impractical for use on gigantic collections with hundreds of
millions of images and thousands of classes. In this paper we show how to utilize
recent results in machine learning to obtain highly efficient approximations
for semi-supervised learning that are linear in the number of images. Specifically,
we use the convergence of the eigenvectors of the normalized graph Laplacian to
eigenfunctions of weighted Laplace-Beltrami operators. Our algorithm enables
us to apply semi-supervised learning to a database of 80 million images gathered
from the Internet.
[[ http://cs.nyu.edu/~fergus/papers/fwt_ssl.pdf | PDF ]]
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips.ppt | PPT ]]
[[ http://cs.nyu.edu/~fergus/code/ssl_eigenfunction.zip | Code ]]
With the advent of the Internet it is now possible to collect hundreds of millions
of images. These images come with varying degrees of label information. “Clean
labels” can be manually obtained on a small fraction, “noisy labels” may be extracted
automatically from surrounding text, while for most images there are no
labels at all. Semi-supervised learning is a principled framework for combining
these different label sources. However, it scales polynomially with the number
of images, making it impractical for use on gigantic collections with hundreds of
millions of images and thousands of classes. In this paper we show how to utilize
recent results in machine learning to obtain highly efficient approximations
for semi-supervised learning that are linear in the number of images. Specifically,
we use the convergence of the eigenvectors of the normalized graph Laplacian to
eigenfunctions of weighted Laplace-Beltrami operators. Our algorithm enables
us to apply semi-supervised learning to a database of 80 million images gathered
from the Internet.
[[ http://cs.nyu.edu/~fergus/papers/fwt_ssl.pdf | PDF ]]
[[ http://cs.nyu.edu/~fergus/presentations/ssl_nips.ppt | PPT ]]
[[ http://cs.nyu.edu/~fergus/code/ssl_eigenfunction.zip | Code ]]
Changed line 3 from:
[[http://cs.nyu.edu/~fergus | Rob Fergus]]
to:
[[http://cs.nyu.edu/~fergus | Rob Fergus]], [[http://www.cs.huji.ac.il/~yweiss/ | Yair Weiss]], [[http://web.mit.edu/torralba/www/ | Antonio Torralba]]
Added lines 1-3:
! Semi-Supervised Learning in Large Image Collections
[[http://cs.nyu.edu/~fergus | Rob Fergus]]
[[http://cs.nyu.edu/~fergus | Rob Fergus]]