Visualizing Concepts: Generative Adversarial Network (GAN) visuals synthesized from semantic vectors
Abstract
We ask what it means to understand a concept in a visual format by integrating knowledge bases into image generation. We source our knowledge from semantic word embeddings that hold the meanings and understanding of words people use in everyday language, and incorporate them into a class of machine learning frameworks known as Generative Adversarial Networks (GANs). We aim to generate visually indeterminate images for concepts that are obscure to imagine, and question the creativity of the machine in doing so.
Advisor
Pasteur, Drew
Department
Art and Art History; Mathematics
Recommended Citation
Issak, Alayt Abraham, "Visualizing Concepts: Generative Adversarial Network (GAN) visuals synthesized from semantic vectors" (2021). Senior Independent Study Theses. Paper 9520.
https://openworks.wooster.edu/independentstudy/9520
Disciplines
Artificial Intelligence and Robotics | Data Science | Interdisciplinary Arts and Media | Other Mathematics | Probability
Keywords
Cognitive Computational Creativity, Generative Adversarial Networks (GAN), Deep Learning, Artificial Intelligence Art
Publication Date
2021
Degree Granted
Bachelor of Arts
Document Type
Senior Independent Study Thesis
© Copyright 2021 Alayt Abraham Issak