In this talk, we will introduce two of my research topics: data-driven PDE identification and image vectorization. For the first part, we will describe what is PDE identification, and how we can identify them from single trajectories perturbed with noise. Most importantly, we will propose some theories about why and when a PDE can be identified. In the second part, we will discuss about image vectorization. We will talk about vector graphics compared to raster images and why do we use them. Then we will describe an effective method that converts raster images to vector graphics based on curve shortening evolution.

学术海报7.pdf