Learning (machine) Learning (2016)


The complex modelling workshop and symposium Crossing Into Physical Realms with Collaborative & Interactive Machine Learning is a three day event exploring and discussing machine learning methods, their application within architecture and the impact of this computational paradigm will have on the ways physical artifacts and spaces are designed, fabricated and behave. The event is structured around a two-day workshop and a one day seminar drawing together key researchers form the fields of architecture, evolutionary robotics, computer science and artificial intelligence. The workshop provides an opportunity for participants to engage with machine learning in a design-led and hands-on way, and the symposium provides an opportunity to discuss, reflect and gain insight from cutting-edge research and practices engaging with machine learning.



The morning session of this first workshop day has provided a theoretical grounding in machine learning concepts, methods of problem representation (direct and in-direct), principle and hybrid paradigms (Evolutionary Computation, Artificial Neural Networks, NeuroEvolution of Augmenting Topologies (NEAT), Compositional Pattern Producing Networks (CPPN) and HyperNEAT), algorithm architecture and application realms. This was delivered by Sebastian Risi (ITU). The afternoon session was led by Robert Vierlinger (University of Applied Arts Vienna and Bollinger+Grohmann) who provided a whistlestop tour of the new Octopus plugin for Grasshopper. Workshop participants followed a series of introductory design exercises using CPPNs – warming up for the design exercise of Day 2 which will see machine learning informed designs fabricated using CNC milling…


Crossing Into Physical Realms with Collaborative & Interactive Machine Learning
is co-organised by CITA and REAL:Robotics, Evolution and Arts Lab, ITU