A heuristic approach to detect novelty data using improved level set methods

The process of mining includes data classification which is one of the most beneficial and constructive methods. As the data is missed during classification process, it if affected on a very large scale and ultimately the mining process is affected. The process of extraction of unknown and new data from a huge dataset which is at times left during classification due to some undefined classification rules is known as novelty detection. We have used various level set methods in order to identify novelty data for temporal data time series. Novelty detection generally focused on the identification of shapes or different patterns, distance and learning based models, density based models, etc. The focus of this paper is to detect and identify the novelty data from the given dataset. The evaluation and experimental results are illustrated at the end. The performance evaluation is done which proves that our proposed method has much higher accuracy than the existing ones.