Criação de representações visuais usando o algoritmo Deepdream e suas implicações conceituais

Abstract

This work intends to explore the aesthetic and creative possibilities of Artificial Neural Networks (RNA), more specifically the Google DeepDream algorithm, and its unfolding in the construction of an artificial consciousness, aiming to point out possible paths of dialogue between human cognitions and the artificial cognitions in digital culture. Through computational tools we seek to expand the creative media of artists and professionals in the media industry, concomitantly with the understanding of the emergence of new languages and aesthetics and their applications. The main goal is to explore the aesthetic and imaging potential of the DeepDream algorithm for creating visual representations of trends topics data from Instagram's social networks, Twitter, and Google's search service. As proof of concept, an experimental prototype of web application was created to generate visual representations through this algorithm. It was intended, therefore, to demonstrate the possibility of using the ANN phenomenon as an aesthetic resource for different areas of human knowledge. In this way, our approach presents the possibility of identifying the emergence of an own aesthetic that dialogues with abstract arts, psychodelism, computational arts, artificial intelligences, large volumes of data, as well as the creative economy and its aspects.