- Úvod
- O nás
- Lidé
- Výzkum
- Knihovna
- Vzdělávání
- Kalendář událostí
- Kariéra
- Kontakty
5.3.2018 14:00 @ Hora Informaticae
In linear regression problems with a large number of parameters to be learned, the classical Ordinary Least Squares (OLS) algorithm presents several drawbacks, such as a possibly infinite number of optimal solutions to the associated optimization problem, and an inadequate interpretability of the resulting vectors of parameters. These issues can be solved by assuming that the vector of parameters generating the data is sparse. In the talk, the well-known Least Absolute Shrinkage and Selection Operator (LASSO) method for sparse regression is reviewed, emphasizing its advantages over OLS for regression problems with high-dimensional vectors of parameters, and discussing issues such as the geomety of the LASSO, algorithms to find its optimal solution, and statistical properties of the LASSO estimator. Recent applications of the LASSO to econometrics problems are also presented.
7.3.2018 10:00 @ Applied Mathematical Logic
14.3.2018 10:00 @ Applied Mathematical Logic