Morphologies and Topologies

Morphologies/Topologies: Bacterially-driven self-organization, pattern-forming and decision-making as models for inference and discovery of new informational relations.

Growth agencies and path attributes 

Growth agencies and their associated path attributes, obtained from an image dataset of pattern-forming bacteria collected by Phylum, are applied to various deep learning generative models. Journeying algorithms are then applied to the dimensionally reduced data in order for the extracted features of bacterial growth to function as “feelers” for exploring potential informational relations between models. These relations can then be strung together into arbitrary manifold linkages that reveal the contextually-dependent nature of their interpretation. So instead of standard linear interpolation through latent space (and therefore areas that have already been deemed closely related) but kind of moving around to different areas of latent space within a model or between two or more models (of almost anything: other images, housing prices, stories/text, etc a kind of cut up technique also doesn’t need to be latente space/generative model. Any data will do (e.g. a database of tweets)