Professor Sebastian Risi receives grant to develop policy learning with neural networks

Professor in the Digital Design Department at the IT University of Copenhagen, Sebastian Risi, has received a grant of approximately 550,000 Danish kroner for his work on neural cellular automata to grow neural network policies capable of adapting to novel complex reinforcement learning tasks.

Sebastian RisiDigital Design DepartmentResearchalgorithmsartificial intelligence

Written 20 April, 2022 12:25 by Theis Duelund Jensen

Professor Sebastian Risi has secured approximately 550,000 Danish kroner from the GoodAI Grants Program to continue his work with neural cellular automata (NCA) to generate 3D structures of varying complexity. Inspired by biological processes, the NCAs were able to regrow parts of simple functional machines significantly expanding the capabilities of simulated morphogenetic systems.

Sebastian Risi, who directs the Creative AI Lab and co-directs the Robotics, Evolution, and Art Lab at ITU, now sets out to grow neural network policies to adapt to different circumstances in complex reinforcement learning tasks in a continuous, self-organized way. Once a policy is developed for the single cell (the update rule), adaptation during the agent’s lifetime will not rely on the more general (but slower) reinforcement learning algorithms or evolutionary algorithms, instead learning by itself through the update rule and knowing from response if it was done correctly.

Cellular automata are complex computational systems governed by a few simple rules. Used to understand how complexity can emerge from a network of identical nodes, cellular automata serve as test beds of more advanced models.

A well-known example of cellular automata is the Game of Life. Invented by British mathematician James Conway in the 1970’s, the Game of Life is a zero-player game. Its evolution is determined by its initial state without input from human players. One interacts with the game by creating an initial configuration and observing how it evolves.

“I’m very happy to receive this grant and to be recognized for the work our group is conducting in this field. I look forward to working on this project and hope to inspire others to investigate approaches at the intersection of deep learning and collective intelligence,” says Sebastian Risi.

The GoodAI Grants Program is part of the GoodAI organization founded in 2014 with the long-term goal of building general artificial intelligence that will automate cognitive processes in science, technology, business, and other fields.

Further information