Exploring Indirect Objectives in Neural Network Training

Kenneth Stanley is a researcher at OpenAI, where he leads the team on Open-endedness. Previously he was a Professor Computer Science at the University of Central Florida, cofounder of Geometric Intelligence, and head of Core AI research at Uber AI labs. His PhD thesis is titled "Efficient Evolution of Neural Networks through Complexification", which he completed on 2004 at the University of Texas. We talk about evolving increasingly complex structures and how this led to the NEAT algorithm that he developed during his PhD. We discuss his research directions related to open-endedness, how the field has changed over time, and how he currently views algorithms that were developed over a decade ago. Episode notes: https://cs.nyu.edu/~welleck/episode9.html Follow the Thesis Review (@thesisreview) and Sean Welleck (@wellecks) on Twitter, and find out more info about the show at https://cs.nyu.edu/~welleck/podcast.html Support The Thesis Review at www.buymeacoffee.com/thesisreview