Let you have two features set X and Y. Using canonical correlation analysis, we get projection weight matrix Wx and Wy, which consists of correlation coefficients. We can project features of X and Y using Wx and Wy in canonical space. Now question is,
How to obtain a single intrinsic feature vector using the projected features in canonical space determined by canonical correlation analysis?
Kandarpa Kumar Sarma
One method is related to canonical correlation analysis (CCA). Another method is called canonical contextual correlation projection (CCCP). Several works on these methods are available. Eg.
http://users.ics.aalto.fi/svirpioj/online-papers/virpioja2011nle_cjo.pdf
http://homepage.tudelft.nl/a9p19/papers/pr_05_cccp.pdf
0 votes
0 thanks
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
Similar questions and discussions