How Does the Curse of Dimensionality Impact Machine Learning?

[zak100]

Homework Statement

What is curse of Dimensionality in the field of Machine Learning and Pattern Recognition?

Homework Equations

No eq just theory

The Attempt at a Solution

Initially the feature space is sparse but as we increase the number of variables, feature space becomes dense. Now we need more computational power for testing those features. Also with more var we have more noise added. This phenomena is called curse of dimensionality. So we have to go for reducing the dimensions which may cause loss of some features.

Is the above correct? What else can i add to it in simple words?

Zulfi.

 

[collinsmark]

Homework Statement

What is curse of Dimensionality in the field of Machine Learning and Pattern Recognition?

Homework Equations

No eq just theory

The Attempt at a Solution

Initially the feature space is sparse but as we increase the number of variables, feature space becomes dense. Now we need more computational power for testing those features. Also with more var we have more noise added. This phenomena is called curse of dimensionality. So we have to go for reducing the dimensions which may cause loss of some features.

Is the above correct? What else can i add to it in simple words?

Zulfi.

Disclaimer: This subject is not my area of expertise.

I think you might have it backwards. As dimensionality increases, the "volume" increases so fast that the available data becomes sparse (not more dense).

I'm not sure if your claim, "with more var we have more noise" is true. You might want to re-think saying that.

You also should bring up how this data's sparseness affects statistical significance.