“For the Sleepers in that Quiet Earth: Experiencing the Behavior of a Deep Learning Neural Network Agent through a Generative Artbook” presented by Audry
Symposium:
Session Title:
- In Between the Cracks (Long Papers)
Presentation Title:
- For the Sleepers in that Quiet Earth: Experiencing the Behavior of a Deep Learning Neural Network Agent through a Generative Artbook
Presenter(s):
Venue(s):
Abstract:
This paper describes an artbook created through the adaptive behavior of a deep learning neural network computational agent as it “reads” a novel. Through this process, the agent builds a model of the syntactic and stylistic principles behind the original text and uses this model to generate new, unforeseen content. A limited set of unique printed copies of the artbook are generated through this process.
Each unique edition of the work thus embodies the learning process of the agent as it goes through the adaptive process, one wherein the agent begins from a state of randomness and gradually refines its output as it reads the novel. I examine the text through an analysis of generated excerpts, discussing how they reveal the behavior of the system as an adaptive agent. Practical and theoretical implications are discussed in the context of generative literature, machine learning, and behavior aesthetics.
