|
Seminarium 28.11.2023
dodano: 24/11/2023
przez: Dariusz Zawisza |
- Referent : Jarosław Duda.
- Tytuł referatu : About decoupling of random variables and multi-feature Granger causality.
- Opis: Wanting to understand relations inside some dataset, a main difficulty is control of complex dependencies between variables. There will be discussed use of conditional CDFs to extract individual information of variables – removing available from the remaining (arXiv:2311.13431). For example to decouple a set of variables: transform into the same number of variables together carrying the same information, but being independent. Or for direct mutual information: without indirect connections. Also for multi-feature Granger causality (arXiv:2305.09478) – enhancing the standard one by evaluation of propagation delay and extracting multiple types of causality.
- Termin: Wtorek 28.11.2023, 12:15.
