PROJECT TITLE :

Active Learning From Imbalanced Data A Solution of Online Weighted Extreme Learning Machine

ABSTRACT:

Active learning is well known for its ability to improve the quality of a classification model while also reducing the complexity of training cases. An uneven data distribution, on the other hand, has been shown in multiple earlier research to easily impair active learning performance. Some existing imbalanced active learning algorithms have limited performance or take a long time to complete. To overcome these issues, this study proposes an effective approach called active online-weighted ELM, which is based on the extreme learning machine (ELM) classification model (AOW-ELM). The following are the main contributions of this paper: 1) the reasons why active learning can be disrupted by an imbalanced instance distribution are discussed in detail, as well as the influencing factors; 2) the hierarchical clustering technique is used to select initially labeled instances in order to avoid the missed cluster effect and cold start phenomenon as much as possible; and 3) the weighted ELM (WELM) is used to select initially labeled instances. The suggested AOW-ELM technique is more effective and efficient than other state-of-the-art active learning algorithms that are specifically built for the class imbalance scenario, according to the experimental findings on 32 binary-class data sets with varied imbalance ratios.