A hybrid model for class noise detection using k-means and classification filtering algorithms

Abstract

Real data may have a considerable amount of noise produced by error in data collection, transmission and storage. The noisy training data set increases the training time and complexity of the induced machine learning model, which led to reduce the overall performance. Identifying noisy instances and then eliminating or correcting them are useful techniques in data mining research. This paper investigates misclassified instances issues and proposes a clustering-based and classification filtering algorithm (CLCF) in noise detection and classification model. It applies the k-means clustering technique for noise detection, and then five different classification filtering algorithms are applied for noise filtering.It also employs two well-known techniques for noise classification, namely, removing and relabeling. To evaluate the performance of the CLCF model, several experiments were conducted on four binary data sets. The proposed technique was found to be successful in classify class noisy instances, which is significantly effective for decision making system in several domains such as medical areas. The results shows that the proposed model led to a significant performance improvement compared with before performing noise filtering.

Applying negative sample denoising and multi-view feature for lncRNA-disease association prediction

“…The noisy training data set increases the training time and complexity of the model. Consequently, identifying noisy instances and then eliminating or correcting them are useful techniques in data mining research ( Nematzadeh et al, 2020 ). Chen et al found that the presence of noisy samples can significantly impact the predictive performance of the LDA model ( Chen et al, 2021b ) Some papers ( Yao et al, 2020 ; Wei et al, 2021 ; Kang et al, 2022 ; Lu and Xie, 2023 ) have used random sampling to create balanced datasets by including an equal number of unknown and positive samples in an attempt to mitigate the impact of unbalanced datasets.…”