English

Hacking Generative Models with Differentiable Network Bending

Computer Vision and Pattern Recognition 2023-12-13 v3 Artificial Intelligence Machine Learning

Abstract

In this work, we propose a method to 'hack' generative models, pushing their outputs away from the original training distribution towards a new objective. We inject a small-scale trainable module between the intermediate layers of the model and train it for a low number of iterations, keeping the rest of the network frozen. The resulting output images display an uncanny quality, given by the tension between the original and new objectives that can be exploited for artistic purposes.

Keywords

Cite

@article{arxiv.2310.04816,
  title  = {Hacking Generative Models with Differentiable Network Bending},
  author = {Giacomo Aldegheri and Alina Rogalska and Ahmed Youssef and Eugenia Iofinova},
  journal= {arXiv preprint arXiv:2310.04816},
  year   = {2023}
}

Comments

12 pages, 10 figures, Machine Learning for Creativity and Design Workshop at NeurIPS 2023