it will be able to label nodes, and to label semantic relations,
but it will be not able to predict semantic relations
Showing posts with label write. Show all posts
Showing posts with label write. Show all posts
2009/04/19
2009/03/25
2009/03/10
2009/02/24
preprocessing samples before give them to the learning machines
There is a large work done on preprocessing samples before give them to the learning machines:
· Remove noise
o Algorithms for detecting noise samples based in knn algorithms.
· Add noise
o Small noise produces a better performance in neural networks (and maybe also in other algorithms).
· Re-structure the dimensionality and distance metrix.
o Nahanalobis distance.
o Scaling the data: It give an improvement in SVM machines
o Kernels: increase dimensions.
o Genetic kernel (GK SVM)
o Removing features:
§ removing dimensions (feature selection)
· information gain (the best)
· mutual information
· x2 statistic chi (second best)
· term strength
§ principal component analysis.
§ neighborhood component analysis
· Re-sampling:
o Under-sampling:
§ Randomly
§ Inconsistent data
§ Duplicate data
§ Removing noise (bis)
o Over-sampling:
§ Randomly
§ SMOTE
§ Border SMOTE-1
§ Border SMOTE-2
§ Adding noise (bis)
§ Give more weight to hard samples.
· Windowed data:
o In some cases context information increases the accuracy.
2009/02/21
Subscribe to:
Posts (Atom)
