This two days research seminar aims at presenting recent advances on the analytical and algebraic perspectives of rough path theory and signatures. Both the theoretical research and their applications will be presented. Applications will include data science, in particular machine learning and neural networks, and financial markets modelling.
Typical topics that will be discussed within the the theory of rough paths and signatures are:
- Hopf and Lie Algebras - and their application toward integration and data analysis
- Statistics - Characterization of data through signatures
- Integration theory and differential equations driven by irregular noise
- Probability theory - Realization of stochastic processes as irregular paths
- Applications to Machine Learning and Finance
Altogether, there will be 12 lectures held by international experts in the field.
The seminar is open for everyone and it is free of charge, however registration is mandatory to allow for appropriate local organisation.
For registration Click Here
The seminar is supported by TMS-Trond Mohn Stiftelsen and STORM Stochastics for Time-Space Risk Models.
-----------------------------------------------------------------------------------------------------------------------
For a preliminary program, see the folder Program in the menu to the left.
For practical information, see the folder Practical information in the menu to the left.
------------------------------------------------------------------------------------------------------------------------
About Rough Paths and Signatures
In the last 20 years, the theory of rough paths has become an active field of research, uniting researchers from probability theory, algebra, and classical analysis to understand the fundamentals of integration and analysis with respect to paths with irregular trajectories.
In applications, the theory of rough paths has been receiving large attention from the communities of machine learning and finance. In machine learning, data arriving as a stream (for example data evolving in time), can be considered as a path. Then the collection of the iterated integrals of a path, which is called the signature, is well used to characterise such data. For example, in a recent study, it has been shown that the signature of a path characterises the law of that path, in a similar way as a probability density function would do. The application of methods and tools from rough paths to the theory of machine learning has provided significant advances in predictive algorithms, also when the objective data is very complex.