Incremental Learning and Knowledge Representation
Parag A. Kulkarni · 2012
This chapter discusses incremental machine learning and knowledge representation. Incremental learning is not just the ability to keep learning from the new data whenever it becomes available, but it is also about testing the hypothesis based on new learning. One approach used for incremental learning is generating an ensemble of learning techniques and classifiers based on the dataset when it becomes available. Unsupervised machine learning is based on similarity and closeness and definitely exhibits a few incremental learning properties. Incremental learning is about responding to new information effectively without retraining. With incremental learning, the learning will take place in a semisupervised way with the existing supervised learning method. The chapter explains the tasks carried out by incremental clustering, which provides a clear idea of how the clustering takes place incrementally. It concludes with a case study on incremental document classification. Controlled Vocabulary Terms knowledge representation; learning (artificial intelligence); unsupervised learning