Neural Network Applications to Cursive Handwriting

David Rumelhart, Stanford Psychology Dept.
der@psych.stanford.edu

Seminar on People, Computers, and Design
Stanford University January 28, 1994

 

The ability to make computers ever smaller and more portable requires the replacement of the keyboard as an input device. There are many possibilites, but hand written inputs, requiring only a pen seems like a good alternative. Unfortunately, printed characters take a relatively long time to write. I have devised a reasonably successful cursive handwriting system. This system, based on neural network technology is promising. I will describe the system we have developed, show its advantages over other methods and discuss future modificaitons to the system to yield better performance.

 

David E. Rumelhart is a professor of psychology, computer science and neuroscience. He has worked in the general areas of Cognitive Psychology, Cognitive Science and Artificial Intelligence. His work has primarily involved the development of computational models of various cognitive phenomena. For the past 15 years his work has focused on the development of "brain style" or connectionis models of cognition and has been interested in the "mind-brain" relationship. This has led to the development of various learning algorithms which have proved useful in the development of a large number of applicaiton areas including: speech recognition, character recognition, process control and a variety of other areas. He is perhaps best know as co-author of the "Parallel Distributed Processing" books.

 

Titles and abstracts for all years are available by year and by speaker.

For more information about HCI at Stanford see