Feature selection (1)
Either the raw or parametric information is described more efficiently, if the feature vectors are projected onto an appropriate subspace.
- In image compression, the Discrete Cosine Transform basis functions are used due to their information packing ability.
When the objective is to approximate the signal, the Karhunen-Loeve transform or principal component analysis (PCA) can yield the basis vectors needed. The features produced by PCA are called most expressive features.