Jump to content

Talk:Nonlinear dimensionality reduction

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia


Linear Methods

[edit]

Why is there a section for linear methods in an article titled non-linear dimensionality reduction? I think it would make more sense to make a new article for linear-dimensionality reduction, and link to it from here. —Preceding unsigned comment added by Headlessplatter (talkcontribs) 16:58, 13 October 2009 (UTC)[reply]

I agree with you. I wanted to move it back to the dimension reduction page, but found that it also needs cleaning up. --Dfalcantara (talk) 22:55, 11 January 2010 (UTC)[reply]

'A' example is non-linear?

[edit]

I'm a little confused about why the first example shows a "nonlinear" reduction in dimensionality. Aren't rotation and scaling linear transformations? Danielx (talk) 06:59, 27 October 2009 (UTC)[reply]

If you read the text carefully (where it says 1024 dimensions), it seems that each "data point" may consist of 1024 values of 0 or 1, and in this case "linear" would be shifts, scales and rotations in this 1024 dimension space. That is, not in the 2 dimensional visual space, which would be a different type of image matching problem. Melcombe (talk) 13:34, 27 October 2009 (UTC)[reply]
Thanks, Melcombe. That makes sense.

Multidimensional Scaling

[edit]

Multidimensional is referred to in the description of some of the techniques, though it is not listed on it's own. Since it is very popular I'm sure it deserves it's own place in the list. T3kcit (talk) 09:22, 4 June 2011 (UTC)[reply]


Completeness of list

[edit]

It seems that t-SNE is currently among the best performing methods, so should it not be introduced? — Preceding unsigned comment added by 129.206.128.114 (talk) 10:05, 17 June 2011 (UTC)[reply]

Thank you, Editors!

[edit]

I offer my deep thanks for everyone who's put effort into this article. As a student of machine learning, it's one of the most succinct and useful pages I've found. Well done! p.r.newman (talk) 18:02, 11 April 2013 (UTC)[reply]

What is nonlinear dimensionality reduction?

[edit]

It takes a long time before the article gets to what nonlinear dimensionality reduction actually is. This should be briefly explained already in the very first sentence, otherwise the reader may have no idea what it is he is actually reading about. —Kri (talk) 23:27, 22 March 2014 (UTC)[reply]

Important Concepts VS Other Algorithms

[edit]

I just glanced over this page but wondered whether there actually is any conceptual difference between the sections "Important Concepts" and "Other Algorithms". if not, I would vote for merging them under one title, which I would suggest being "(List of) Methods", "(List of) Algorithms", or something like that. — Preceding unsigned comment added by 81.191.194.157 (talk) 08:09, 21 October 2020 (UTC)[reply]