r15 - 23 Oct 2012 - 15:52:42 - FredericBastienYou are here: TWiki >  Public Web  > MnistVariations

Variations on the MNIST digits

We are interested in creating simple variations on the popular MNIST digits so as to introduce controllable factors of variation. The generative process used to generate the datasets is the following:

  • Pick sample $(x,y) \in {\cal X}$ from the MNIST digit recognition dataset (http://yann.lecun.com/exdb/mnist/);
  • Create a perturbed version $\widehat{x}$ of $x$ according to some factors of variation;
  • Add $(\widehat{x},y)$ to new dataset $\widehat{{\cal X}}$;
  • Go back to 1 until enough samples are generated.

Introducing multiple factors of variation leads to the following benchmarks:

  • mnist-rot: the digits were rotated by an angle generated uniformly between 0 and $2 \pi$ radians. Thus the factors of variation are the rotation angle and the factors of variation already contained in MNIST, such as handwriting style:

rotation_examples.png

  • mnist-back-rand: a random background was inserted in the digit image. Each pixel value of the background was generated uniformly between 0 and 255;

mnist_back_random.png

  • mnist-back-image: a patch from a black and white image was used as the background for the digit image. The patches were extracted randomly from a set of 20 images downloaded from the internet. Patches which had low pixel variance (i.e. contained little texture) were ignored;

mnist_back_image.png

  • mnist-rot-back-image: the perturbations used in mnist-rot and mnist-back-image were combined.

mnist_rot_back_image.png

Download links

All the datasets are provided as zip archives. Each archive contains two files -- a training (and validation) set and a test set. We used the last 2000 examples of the training sets as validation sets and, in the case of SVMs, retrained the models with the entire set after choosing the optimal parameters on these validation sets. Data is stored at one example per row, the features being space-separated. There are 784 features per example (=28*28 images), corresponding to the first 784 columns of each row. The last column is the label, which is 0 to 9.

Link to dataset Size File description
MNIST basic 23M packed; 599M + 144M unpacked 12000 train, 50000 test
MNIST + background images 88M packed; 599M + 144M unpacked 12000 train, 50000 test
MNIST + random background 219M packed; 599M + 144M unpacked 12000 train, 50000 test
Rotated MNIST digits 56M packed; 338M + 56M unpacked 12000 train, 50000 test
Old version: Rotated MNIST digits 284M packed; 599M + 144M unpacked 12000 train, 50000 test
Rotated MNIST digits + background images 115M packed; 599M + 144M unpacked 12000 train, 50000 test
Old version: Rotated MNIST digits + background images 124M packed; 599M + 144M unpacked 12000 train, 50000 test

Contact

Please contact us if you are interested in the (MATLAB) scripts that we used to generate this data.

-- DumitruErhan - 21 Jun 2007

toggleopenShow attachmentstogglecloseHide attachments
Topic attachments
I Attachment Action Size Date Who Comment
pngpng mnist_back_image.png manage 10.0 K 21 Jun 2007 - 11:06 DumitruErhan MNIST background images
pngpng mnist_back_random.png manage 13.4 K 21 Jun 2007 - 11:06 DumitruErhan MNIST background random
pngpng mnist_rot_back_image.png manage 9.9 K 21 Jun 2007 - 11:07 DumitruErhan MNIST rotation + background images
pngpng rotation_examples.png manage 5.0 K 21 Jun 2007 - 11:07 DumitruErhan MNIST rotation
Edit | WYSIWYG | Attach | Printable | Raw View | Backlinks: Web, All Webs | History: r15 < r14 < r13 < r12 < r11 | More topic actions
 
Home
This site is powered by the TWiki collaboration platformCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback