Generating Novel Glyph without Human Data by Learning to Communicate
This addresses the challenge of creating visual symbols from scratch, potentially offering insights into the origins of writing systems, but it appears incremental as it builds on existing communication-based frameworks.
The paper tackles the problem of generating novel glyphs without human data by training a generator and classifier to communicate via visual symbols, resulting in glyphs that resemble human-made ones.
In this paper, we present Neural Glyph, a system that generates novel glyph without any training data. The generator and the classifier are trained to communicate via visual symbols as a medium, which enforces the generator to come up with a set of distinctive symbols. Our method results in glyphs that resemble the human-made glyphs, which may imply that the visual appearances of existing glyphs can be attributed to constraints of communication via writing. Important tricks that enable this framework are described and the code is made available.