In prediction model, the pre-processing has major effect before do binary classification. For selecting feature, feature selection technique is able to applied on pre-processing step. Recently, the researcher did some hybrid feature selection techniques. Are there any recent and novel feature selection techniques, it used to select the features? Thanks
Dear Daniel
Have you looked at CMI?
www.jmlr.org/papers/volume5/fleuret04a/fleuret04a.pdf
Best wishes
Georgios
Hi Daniel,
have you considered a Monte Carlo-based LDA method? Basics on its application for feature selection and experimental results are described in the attached papers.
Cheers,
Uwe
Conference Paper Sensorless drive diagnosis using automated feature extractio...
Conference Paper Feature Extraction and Reduction Applied to Sensorless Drive Diagnosis
u can try genetic algorithms or bio-inspired algorithms for feature selection
You could also use sparsity-inducing techniques (e.g., LASSO, Elastic-net, ...).
The main advantage is that the feature selection occurs during the learning process, not as a pre-processing step.
May be, this will be helpful.
Article Quality assessment of data discrimination using self-organizing maps
See similar discussions:
https://www.researchgate.net/post/How_can_I_select_optimal_features_from_feature_selection_method/1
https://www.researchgate.net/post/What_is_the_best_way_to_identify_the_most_relevant_sufficient_features_for_classification2
You can find simple and effective method of FS (wrapper) in my article:
https://www.researchgate.net/publication/262311633_Tournament_feature_selection_with_directed_mutations
Conference Paper Tournament Feature Selection with Directed Mutations
The most appropriated Feature Selection technique depends on your data (quality, amount and other things matter on feature selection performance).
As I don't have details of your field and available data, I would suggest you start by evaluating Feature Selectors implemented in WEKA:
http://www.cs.waikato.ac.nz/ml/weka/
They are general techniques that work well in various domains, and one (or some) of them may be your best option.
I will add a few words to Felipe Leno da Silva.
I think we should search for the techniques that are the most suitable to our data. There is no unique feature selection method which suits for all needs. Therefore, you should also check "old" methods. WEKA supports many such tecniques under attribute selection tab. As a last point, with the use of wrappers and search strategies together, one can find nearly endless hybrid feature selection techniques.
regards
As Shaveta Chutani says, you can use GE for feature selection and also for feature construction. What about LASSO for feature selection?
Bests!
More than often, some researchers do use the terms feature selection and feature extraction interchangeably but they are different. I wish to be sure which you are really talking about.
hi ,
Please go through the following links
http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6620083&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D6620083
http://matwbn.icm.edu.pl/ksiazki/amc/amc11/amc1131.pdf
https://www.lri.fr/~pierres/donn%E9es/save/these/articles/a-rough-sets-based.pdf
Hi sir,
I have a paper for you:"infinite feature selection" ICCV 2015. This technique is able to deal with huge amount of features and it will perform the rank of your variables / features. Then one can select a subset starting from the outcome of the method. It ranks the features by importance. The source code for matlab is available on the project page. http://giorgioroffo.it/projects/infFS/
- Badges
- Science topic
- Similar topics
- Data Mining
- Classification