Index
Focus
Neural Walks
❋ ✦ ✜
Neural Walks is an attempt to understand how a complex computer system operates and how we might experience it as humans. The project rests on the foundation of the core methods of computer vision and machine learning. The artistic statement of the project is shaped around meta-level curiosity: neural networks are one of the most powerful computational tools, however, processing billions of images and generating billions of new ones, in the paradigm of computer vision all they ‘see’ - are grids of kernels, noise, and RGB values. Playfully, the project offers an experience of being in a body of the neural network but seeing as a human. The players are invited on a journey through the network of images where the algorithm computes a multitude of pathways between them based on their visual features and metadata. Along the walk, viewers can observe peculiar behavior of the neural network in action, e.g., style transfer from one image to another. As they traverse through, the grids of images loose their rigid divisions. Not only textures, but also objects start to blend, turning the image set into a homogeneous otherworldly landscape.
PROJECT STAGE: Proposal