Making new games requires a large amount of design and technical knowledge, limiting who can make games and for what purpose. This project looks to decrease the time and knowledge requirements of game design through machine learning. This project received a Best Paper Award at the International Conference on Computational Creativity, and coverage in the BBC, Rolling Stone, and WIRED.BBC Rolling Stone WIRED
Machine learning has the potential to support designers, helping creatives make better work faster. In this project I focus on building tools for designers, like an intelligent game level editor with an intelligent, active learning assistant, ML approaches for predicting user experience, and tools to help in visual theming. Due to this work I received a 2018 Unity Graduate Fellowship and coverage in The Register.The Register YouTube: Active Learning Co-Creative Level Editor
Computational creativity is the field of research on computationally representing human creativity. I believe that the application of these computational creativity methods could allow machine learning to move from predicting what has come before to anticipating and creating new possibilities. In this work I applied computational creativity to create image classification and generation in a transfer learning framework (beating state of the art baselines), and built a benchmark for the development of new creative ML agents.