Differentiable optimisation for constrained three-dimensional shape design. Deep learning has revolutionized many areas of machine learning and now stands ready to address problems in science and engi

Description

Differentiable optimisation for constrained three-dimensional shape design. Deep learning has revolutionized many areas of machine learning and now stands ready to address problems in science and engineering. This project will apply techniques from deep learning, specifically the ability to solve complex optimisation problems, to the task of constrained three-dimensional shape design. The outcome of the project will be a set of mathematical tools to assist engineers and designers. Such tools will accelerate the design and evaluation of three-dimensional shapes, such as airplane wings and implantable medical devices, that optimise operational performance while meeting manufacturing and structural constraints, significantly reducing development costs and time-to-market.. Scheme: Discovery Projects. Field: 4611 - Machine Learning. Lead: Prof Stephen Gould