Talk:Feature extraction

Arcticle start
This is the first complex article that I've started, and I've got an awful feeling that I'm going to be putting far too much into it - so I'll take advice as an when. The topic itself is fairly technical, so it'll take a while to update as I'll have to decide what kind of level I want to pitch it at (by looking at other similar articles), and then get some of the mathematics together.

Ahhh-I've taken too much on :c>

--

Sarcas

Wikification
As well as a bit of 'wikification', I've turned a couple of the headings here into links, because I think they deserve their own article. Perhaps this page would best serve the role of linking these areas, and introducing each one briefly. Of course, you're perfectly entitled to disagree with this judgement, and merge the articles back. My main aim was to write an article on the Hough transform, because I just implemented it for a piece of coursework :D - IMSoP 17:46, 18 Jan 2004 (UTC)

Article is too biased
This article is too biased in the direction of image processing. Please adapt that into the broader category of dimensionality reduction and draw a clear line with respect to shape recognition:
 * JMLR Special Issue on Variable and Feature Selection

Please add also more references to allow unexperienced users to contribute. JKW 09:56, 23 April 2006 (UTC)

I've added a link to a new page I created on Corner detection. May be the bias on image processing can be reducing by citing a few speech processing examples? I might may some speech processing pages if time permits, will link them to here when done. Retardo 02:50, 29 April 2006 (UTC)

The article is also too general
This article appears to have an extremely general scope. It aims at too much and the result does not at all match the expectations. Tpl

Feature Selection vs. Extraction
So what is the difference between feature selection and feature extraction? Should these two articles be merged? — Preceding unsigned comment added by 173.3.109.197 (talk) 17:18, 21 April 2012 (UTC)

Relevant component analysis and Distance Metric Learning?
Would it be appropriate to briefly mention RCA (relevant component analysis, more on that here: http://www.wikicoursenote.com/wiki/Relevant_Component_Analysis) and distance metric learning in this article?

It is one perspective on feature extraction - minimize the variability in the data which is not correlated to the task at hand. I've not looked up any good sources, but I was reading a paper from an academic journal [jmlr.csail.mit.edu/papers/volume13/ying12a/ying12a.pdf that compared their distance metric learning with RCA]. I think RCA and distance metric learning should have their own pages, but I was thinking until they could get started here. — Preceding unsigned comment added by 150.135.222.166 (talk) 23:54, 27 September 2012 (UTC)

Image processing is tangential
This is a case where my splitter side runs out of patience. Lumping "image processing" in the same page with feature extraction in a generic machine learning / statistics context helps no one.

Would not feature extraction (image processing) not better serve both masters? &mdash; MaxEnt 01:16, 16 January 2017 (UTC)

Online book
http://www.feat.engineering/

Feature Engineering and Selection: A Practical Approach for Predictive Models Max Kuhn and Kjell Johnson

2019-06-21 The goal of our previous work, Applied Predictive Modeling, was to elucidate a framework for constructing models that generate accurate predictions for future, yet-to-be-seen data. This framework includes pre-processing the data, splitting the data into training and testing sets, selecting an approach for identifying optimal tuning parameters, building models, and estimating predictive performance. This approach protects from overfitting to the training data and helps models to identify truly predictive patterns that are generalizable to future data, thus enabling good predictions for that data. Authors and modelers have successfully used this framework to derive models that have won Kaggle competitions (Raimondi 2010), have been implemented in diagnostic tools (Jahani and Mahdavi 2016; Luo 2016), are being used as the backbone of investment algorithms (Stanković, Marković, and Stojanović 2015), and are being used as a screening tool to assess the safety of new pharmaceutical products (Thomson et al. 2011). — Preceding unsigned comment added by 66.35.36.132 (talk) 20:29, 6 August 2020 (UTC)