Little 1200

This dataset was collected at Bell Laboratories in 1988 by Isabelle Guyon and Wayne Hubbard. It includes 10 sets of handwritten digits 0-9, written by 12 writers. It was digitized as 16x16 binary images.

It was used to benchmark machine learning algorithms in the PhD thesis of Isabelle Guyon, whose results were published in:

Guyon, I., Poujaud, I., Personnaz, L., Dreyfus, G., Denker, J., & Le Cun, Y. (1989). Comparing different neural network architectures for classifying handwritten digits. In International 1989 Joint Conference on Neural Networks (pp. 127-132). IEEE.

You may cite this paper [BibTex] if you use this dataset, which can be downloaded from [Here].