For my spring research seminar class I carried out a semester-long project studying signature methods and signature kernels along with their applications in machine learning tasks. I gave multiple presentations and wrote a cumulative research report detailing my work and findings.
The signature method is a recent development from the field of rough path theory which relies on iterated integrals of path data. They have recently been used in machine learning tasks with great success, making them a powerful tool for dealing with time series. They are able to powerfully capture information about stream like data and can be paired with tools such as Kernels for enhanced learning.
My Research Report, along with one of the presentations I gave on the materials, is presented below.
Link: Research Report
Link: Presentation Slides