Deconstructing Representation

In Deconstructing Representation machine learning is used as an explorative approach to visualise patterns within a dataset of images. Compiling a dataset of Instagram images saved by the artist over time and training a generative adversarial network (GAN) to produce images reminiscent of that dataset, the resulting images show the gradual buildup and breakdown of computational representations and visual affinities.

Source: http://www.rosemary-lee.com/Deconstructing-Representation