Lecture 15 - Dimensionality Reduction

Question 1

What is dimensionality reduction?

Answer 1

It is an unsupervised approach that transforms the feature vectors into a lower dimensional space.

Question 2

Are features preserved in dimensionality reduction?

Answer 2

No are transformed into a smaller feature vector. It is considered a compression.

Question 3

Are features preserved with feature selection?

Answer 3

Yes

Question 4

List the types of dimensionality reduction.

Answer 4

PCA (linear)
Locally linerar embedding
t-SNE
MDS

Question 5

What is PCE and how does it work?

Answer 5

Principal component analysis looks for a combination of features that captures the variance of the original features.

Question 6

What are manifold learning techniques?

Answer 6

They are a subset of nonlinear models of dimensionality reduction.