A crucial task in Physics is the extraction of degrees of freedom in a dynamical system. Recently, a neural network architecture was created to perform exactly this task. The architecture is called a β-Variational Auto-Encoder (β-VAE) and is illustrated the figure below. The original paper applied this to planetery orbits....
[Read More]
How to shape an elephant
Neural networks for the design of photonic components
In the field of Photonics, optical components are created based on experience and domain knowledge. Creating complex structures with a high-dimensional design space is therefore often difficult. To overcome this problem, deep learning has been introduced in recent years. The neural networks of deep learning are applied in what we...
[Read More]
How to segment the brain
Convolutional neural networks for MRI image analysis
A crucial task in medicine is the analysis of medical images. This is traditionally done by radiologists. With the advent of better imaging techniques, there is more and more data created on each patient, which also means more data to be analyzed by radiologists. To reduce their workload and assist...
[Read More]