Deep learning describes abstract paintings

Deep Learning describes Abstract Paintings is a book that attempts to put in relation said-iconic abstract paintings and captions of these that have been generated by advanced image recognition algorithms.

For this project, our interest was lying mainly between differences of perception and language shifts between a generally accepted conception of what an abstract painting is (a non-figurative work of art that wishes to resist interpretation – an aura and a context that transcends the work of art) and a mode of interpretation that tries to be generic, objective and absolute. 

 Deep Learning type algorithms are algorithms that attempt to replicate the human learning process. They are generally precise when it comes to “figurative” images (to detect human faces) and can generate sentences with an advanced syntax level. For a so-called “abstract” image, it is quite different; results can be more thought of like a guessing game, or a form of pareidolia (the phenomenon of somebody perceiving identifiable forms out of random shapes, i.e.cloud evoking objects). The generated material was laid out a bit like a poetry book, where content, coming from “artificial intelligence”, coexists with the images of the selected works.

 Anteism Books Publisher