Regularization of restricted canonical correlation analysis

Canonical correlation methods are a family of statistical methods. The original canonical correlation analysis was developed to examine linear relationships between two sets of variables. In order to increase the flexibility of the original method, several extensions of canonical correlation analysis have been proposed. Two extensions will be discussed, restricted canonical correlation analysis and restricted kernel canonical correlation analysis. The standard technique for solving canonical correlation analysis problems is based on an eigenvalue problem. The search process has an exponential time complexity and even problems with a few tens of variables cannot be solved in a feasible time. In this lecture an alternative technique will be described, so that even problems with tens of thousands of variables can be solved in a feasible time. The problem of cross-language information retrieval is shown as an example.

Link to the videoconference:

Topic: IBMI predavanje (nov 26)
Time: Nov 26, 2020 12:00 PM Budapest

Join Zoom Meeting
https://uni-lj-si.zoom.us/j/92726738578?pwd=amc2WUsrSWlYcE5mcDVmRjBCVzFsdz09

Meeting ID: 927 2673 8578
Passcode: ibmi