IEEE WIECON-ECE 2018 Novel Class Detection in Concept Drifting Data Streams Using Decision Tree Leaves
| Autor: | Akash Sarkar, Famina Alam, Mozzammel Haque, Deepita Saha, Swakkhar Shatabda, Chowdhury Mofizur Rahman, Dewan Md. Farid |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Data stream
Class (computer programming) Concept drift Computer science Data stream mining business.industry Big data Weather forecasting Decision tree 02 engineering and technology Intrusion detection system computer.software_genre 020204 information systems 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Data mining business computer |
| Zdroj: | 2018 IEEE International WIE Conference on Electrical and Computer Engineering (WIECON-ECE). |
| DOI: | 10.1109/wiecon-ece.2018.8782911 |
| Popis: | Concept drifting data streams often occurs in weather forecasting, intrusion detection and other applications. One of the difficulties with handling concept drifting data streams is the existence of novel classes in the data stream that arrives after the training of the model on the existing class instances. In this paper, we present a novel class detection algorithm in concept based on the instance distribution in the decision tree leaves. Our proposed algorithm is easy to implement and use compared to complex ensemble based methods. We have tested the performance of our algorithm on several datasets and it shows significantly improved results compared to previous state-of-the-art algorithm using standard evaluation methods and metrics. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|