Speaker
Prof.
Wayne de Fremery
(Sogang University)
Description
This talk will suggest that a form of deep learning known as Generative Adversarial Networks (GANs) can be usefully incorporated into bibliographical investigations of older East Asian texts. It will do so by demonstrating that GANs can be used to generate historically accurate representations of Korean, Japanese, and Chinese xylographyic and typographic shapes. To demonstrate their usefulness to historical bibliography, the talk will focus on a description of how GANs have been used to generate nearly all of the unique characters likely to appear in an important facsimile of the Qisha Canon, thereby speeding up its transcription and advancing work underway in the Buddhist community to transcribe all of the documents associated with the Chinese Buddhist Canon into machine-readable text.
