This is an old revision of the document!
This early draft of the graph transformer network work was written in April 2015. I've decided to offer it online because, in retrospect, this is a much better document than any of the published papers.
Abstract: We present here a general architecture for building Automatic Document Analysis Systems. This architecture is composed of a succession of modules transforming graphs describing lower-level hypotheses on the documents into graphs describing higher level hypotheses. This architecture generalizes techniques used in Neural Networks, Optical Character Recognition, Natural Language Processing and Speech Recognition.
transducer-1996.djvu transducer-1996.pdf transducer-1996.ps.gz
@misc{bottou-1996, author = {Bottou, L\'{e}on and Bengio, Yoshua and {LeCun}, Yann}, title = {Draft report: Document Analysis with Transducers}, year = {1996}, month = {July}, optnote = {Made available on http://leon.bottou.com/publications in 2015}, url = {http://leon.bottou.org/papers/bottou-1996}, }