“Diaspora Letters” is an automatic computational asemic writing project which uses machine learning to abstract hand writing letters by extracting shapes and patterns from strokes with ignoring meaning of them in various alphabets.
Human can recognise what language it is without understanding meaning when he/she hears someone’s conversation on street, if he/she have heard of the language. it means human learning sound without meaning from language. Will it happen to letters too?
This project uses same algorithm from “SDM2 – Letters”, however machine learning program learns separately for each various alphabets. Therefore the purpose is to extract shapes of strokes from them.
Using hand-writing letters of Arabic, Chinese, Cyrillic, Hebrew and Japanese.
branch test project of “SDM2-Letters” (collaboration with Takahiro YAMAGUCHI)
generated strokes links
Machine Learning Programming: Hironori Sakamoto
photo: Ikumi Hirata
Shown in HUMAN ♥ ROBOT at SPEKTRUM
*asemic writing (from Wikipedia)
Asemic writing is a wordless open semantic form of writing. The word asemic means “having no specific semantic content”.