Introduction
Machine learning offers a new and universal way of looking at the world, which is radically different than traditional domain-focused and theory-based approaches. However, while new machine learning algorithms such as deep learning have created a wave of applications in computer vision and natural language translation, there are a lot of complex applications which are yet to be investigated.
In recent years there have been significant advances in reinforcement learning, generative models and topological data analysis for media creation, design exploration, and urban analytics.
About the event
This talk will focus on the domain of machine learning for creativity and design, looking at the applications of unsupervised learning in the context of urban system modelling, structural design explorations, and the use of reinforcement learning in generative arts which provides a glimpse of how machine learning may one day find its place in the toolset of artists and designers.
Register now
Registration is now closed. This event is sold out.
Agenda
15:00 - 15:15 Registration
15:15 - 15:20 Introduction and welcome
15:20 - 16:00 Machine learning literacy for designers and engineers (Vahid Moosavi, CAAD, ETH)
16:00 - 16:20 Discussion
16:20 - 17:00 Unsupervised doodling and painting with generative agents (S. M. Ali Eslami, Google DeepMind)
17:00 - 17:20 Discussion
17:20 - 18:00 Networking reception
Talk abstracts
Dr Vahid Moosavi (Senior Researcher, CAAD, ETH)
Machine learning literacy for designers and engineers
Machine learning and data together offer a universal way of looking at the world phenomena, which is radically different than the classical expert-based disciplinary research. This new approach of computational modeling has inverted the classical notion of expertise from “having the answers to the known questions” to “learning to ask good questions”. This will have important consequences on the ways we think about academic research and education.
In this talk, while presenting a range of projects from different fields such as structural design, urban flood risk modeling, planetary urban analysis and real estate market I provide a “generic typology” of machine learning applications across domains. Finally, I would like to point out to those application areas in engineering that the use of generic machine learning is a game changer and those areas of design that machine learning can work only in a personal manner and in a co-existence with machine learning literate humans.
Dr S. M. Ali Eslami (Research Scientist, Google DeepMind)
Unsupervised Doodling and Painting with Generative Agents
The computational techniques that drive the field of machine learning are increasingly being used for creative endeavours. In particular, recent advances in generative modelling provide a tantalising glimpse of how machine learning may one day find a place in the modern artist's tool set. In this talk I will investigate using reinforcement learning agents as generative models of images. A generative agent controls a simulated painting environment, and is trained with rewards provided by a discriminator network simultaneously trained to assess the realism of the agent's samples. We find that when sufficiently constrained, generative agents can learn to produce images with a degree of visual abstraction, despite having only ever seen real photographs and no trajectories. And given enough time with the painting environment, they can produce images with considerable realism. These results show that, under the right circumstances, some aspects of human drawing can emerge from simulated embodiment, without the need for external supervision, imitation or social cues.