A platform for research: civil engineering, architecture and urbanism
Urban Portraits
Machine learning (ML) has been extensively used since 2015 for image style transfer, with particular popularity of neural networks trained on collections of images from early‐twentieth‐century modern painters and earlier. The nature of the training process for neural networks results in the biases present in the original data set to reflect on the results a particular model produces. However, one of the most insightful examinations of creative ML might be the use of this bias to explore alternative representations outside the main domain a model was generated from. A neural network trained on Kandinsky's geometrical compositions may style an urban cityscape with patterns never produced by the author's hand. It is up for discussion whether we are to believe that the polynomial weighting performed by fitting a neural network can shed some light about the nature of human art.
Urban Portraits
Machine learning (ML) has been extensively used since 2015 for image style transfer, with particular popularity of neural networks trained on collections of images from early‐twentieth‐century modern painters and earlier. The nature of the training process for neural networks results in the biases present in the original data set to reflect on the results a particular model produces. However, one of the most insightful examinations of creative ML might be the use of this bias to explore alternative representations outside the main domain a model was generated from. A neural network trained on Kandinsky's geometrical compositions may style an urban cityscape with patterns never produced by the author's hand. It is up for discussion whether we are to believe that the polynomial weighting performed by fitting a neural network can shed some light about the nature of human art.
Urban Portraits
Carta, Silvio (editor) / del Castillo y López, Jose Luis García (author)
Machine Learning and the City ; 501-505
2022-05-21
5 pages
Article/Chapter (Book)
Electronic Resource
English
Online Contents | 2014
Online Contents | 2014
TIBKAT | 2022
|Living in the city : urban interiors and portraits
TIBKAT | 2019
|Online Contents | 2014