Confidence in prediction: an approach for dynamic weighted ensemble.

Do, Duc Thuan; Nguyen, Tien Thanh; Nguyen, The Trung; Luong, Anh Vu; Liew, Alan Wee-Chung; McCall, John

Abstract

Combining classifiers in an ensemble is beneficial in achieving better prediction than using a single classifier. Furthermore, each classifier can be associated with a weight in the aggregation to boost the performance of the ensemble system. In this work, we propose a novel dynamic weighted ensemble method. Based on the observation that each classifier provides a different level of confidence in its prediction, we propose to encode the level of confidence of a classifier by associating with each classifier a credibility threshold, computed from the entire training set by minimizing the entropy loss function with the mini-batch gradient descent method. On each test sample, we measure the confidence of each classifier’s output and then compare it to the credibility threshold to determine whether a classifier should be attended in the aggregation. If the condition is satisfied, the confidence level and credibility threshold are used to compute the weight of contribution of the classifier in the aggregation. By this way, we are not only considering the presence but also the contribution of each classifier based on the confidence in its prediction on each test sample. The experiments conducted on a number of datasets show that the proposed method is better than some benchmark algorithms including a non-weighted ensemble method, two dynamic ensemble selection methods, and two Boosting methods.

Citation

DO D.T., NGUYEN T.T., NGUYEN T.T., LUONG A.V., LIEW A.W.-C. and MCCALL J. 2020. Confidence in prediction: an approach for dynamic weighted ensemble. In Nguyen N., Jearanaitanakij K., Selamat A., Trawiński B. and Chittayasothorn S. (eds.) Intelligent information and database systems: proceedings of the 12th Asian intelligent information and database systems conference (ACIIDS 2020), 23-26 March 2020, Phuket, Thailand. Lecture Notes in Computer Science, 12033. Cham: Springer [online], part 1, pages 358-370. Available from: https://doi.org/10.1007/978-3-030-41964-6_31

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 12th Asian intelligent information and database systems conference (ACIIDS 2020)
• Start Date: Mar 23, 2020
• End Date: Mar 26, 2020
• Acceptance Date: Dec 1, 2019
• Online Publication Date: Mar 4, 2020
• Publication Date: Mar 31, 2020
• Deposit Date: Apr 9, 2020
• Publicly Available Date: Apr 9, 2020
• Publisher: Springer
• Peer Reviewed: Peer Reviewed
• Volume: 12033
• Pages: 358-370
• Series Title: Lecture notes in computer science
• Series ISSN: 0302-9743
• Book Title: Intelligent information and database systems
• ISBN: 9783030419639
• DOI: https://doi.org/10.1007/978-3-030-41964-6_31
• Keywords: Supervised learning; Classification; Ensemble method; Ensemble learning; Multiple classifier system; Weighted ensemble
• Public URL: https://rgu-repository.worktribe.com/output/891647