A REVIEW ON INCREMENTAL MACHINE LEARNING METHODS, APPLICATIONS AND OPEN CHALLENGES

In recent years, Incremental learning has gained significant attention in the context of large-scale data and stream learning. It learns the continuously arriving data streams over time with the utilization of limited memory without compromising the accuracy of the system. Incremental learning can continuously learn the knowledge from the new data and also capable of maintaining the previously learned knowledge. It plays significant role in real-world applications that learn the arrival of data over time in an ever-changing environment. Hence, this work formalizes the characteristics of continuously arriving data and the concept of incremental learning. It also reviews the conventional approaches with the theoretical foundations and discusses the challenges encountered by incremental learning approaches. This survey examines the popular machine learning and deep learning algorithms that support incremental learning. Finally, it also provides research challenges and future research opportunities for incremental learning models and future research opportunities for incremental learning models.