Austin Dill, Eunsu Kang, Songwei Ge, Barnabas Poczos

toiletdumpling_featured

How to Make a Dumpling with Toilet (2019). This video work explores how machine learning algorithms view objects, space, and sound by visualizing transformations learned by a generative algorithm  set to the sound of audio generated by a second AI agent. This work shows that AI can find the connections between mundane objects that are imperceptible to humans in addition to showing the potential of machine learning for 3D art. 

The machine learning techniques used to generate this include methods discussed in the submitted paper “Learned Interpolation for 3D Generation” for the visualized point clouds as well as a pretrained WaveGAN model for sound generation