Users with special circumstances, such as limited mobility or physical strength, are often unable to utilize the normal keyboard of a device. To overcome these difficulties, these users utilize alternative mechanisms for typed input, such as a mouse, trackpad, switches, buttons, etc. These mechanisms operate by mapping the full set of possible inputs onto a limited number of buttons, which makes their use cumbersome and slow. This disclosure utilizes Huffman coding to optimize the encoding of a large set of symbols into a set of codewords based on the probability of use of each symbol, calculated via a trained machine learning model. Given a reasonably accurate machine-learned prediction model, the techniques of this disclosure ensure that generating the desired typed input can be accomplished with minimal number of switch selections.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Membrives, Etienne J., "User Interface for Input With Switches Using Machine Learned Huffman Codes", Technical Disclosure Commons, (August 14, 2018)