Further more, the ADASYN algorithm can also be modified
to facilitate incremental learning applications. Most current
imbalanced learning algorithms assume that representative
data samples are available during the training process. However,
in many real-world applications such as mobile sensor
networks, Web mining, surveillance, homeland security, and
communication networks, training data may continuously become
available in small chunks over a period of time. In
this situation, a learning algorithm should have the capability
to accumulate previous experience and use this knowledge
to learn additional new information to aid prediction and
future decision-making processes. The ADASYN algorithm
can potentially be adapted to such an incremental learning
scenario. To do this, one will need to dynamically update the ri
distribution whenever a new chunk of data samples is received.
This can be accomplished by an online learning and evaluation
process.