Volume & Issue no: Volume 3, Issue 10, October 2014
_____________________________________________________________________________
Title: |
A survey on Oversampling Techniques for Imbalanced Learning |
Author Name: |
Miss. Reshma K. Dhurjad, Prof. Mr. S. S. Banait |
Abstract: |
ABSTRACT
In machine learning and data mining data imbalance is a key source of performance degradation. Key reason behind this
degradation is that all available algorithms assume a balanced class distribution for learning. In many real-world applications,
the data available for learning are highly imbalanced. Imbalanced data means where one class severely out-represent another
class. In these scenarios, the learning algorithms tend to bias toward the less important negative class or majority class with
larger instances. Although, there is no single best technique to deal with imbalance problems, sampling techniques have been
shown to be very successful in recent years. To address imbalanced learning issue oversampling of minority class is done.
There are various Oversampling techniques which can be used to reestablish the class balance. Oversampling method is a data
level method. The main advantage of data level methods is that they are self-sufficient. The methods at data level modify the
distribution of the imbalanced datasets, and then these modified i.e. balanced datasets are provided to the algorithm to improve
the Imbalanced learning.
Keywords:- Classification, Imbalanced data, learning, oversampling |
Cite this article: |
Miss. Reshma K. Dhurjad, Prof. Mr. S. S. Banait , "
A survey on Oversampling Techniques for Imbalanced Learning" , International Journal of Application or Innovation in Engineering & Management (IJAIEM) ,
Volume 3, Issue 10, October 2014 , pp.
279-284 , ISSN 2319 - 4847.
|
Full Text [PDF] Back to Current Issue  |
|
