demension reduction
Table of Contents
technique to map a set of 120-vectors to 3-vectors (or in jargons, 120 feature observation to 3 feature observation) set of [5 4 3 4 4 6 4 2 12 3 5 7 4 3 2 1]s -> set of [1 0.5 3]s
- useful/moral
- meaningful demension
- there are only that much meaningful demensions
- manifold
- if you map a 2-vector set to a 3-vector set, they’ll form a manifold in the 3-vector space.
- efficiency/complexity
- obviously the less demensions the more efficiency and less complexity.
- techniques
- PCA - principle component analysis: iteratively compute most significant component
- - cluster features to form
Backlinks
A demension reduction idea like PCA, NMF.
Group Features into sementic groups which is represented with a matrix. With the matrix, transform the High-dimensional vector into a lower-dimensional vector.