Adaptive Particle Swarm Optimization Algorithm Ensemble Model Applied to Classification of Unbalanced Data

Abstract

Unbalanced data classification is a major challenge in the field of data mining. Random forest, as an ensemble learning method, is usually used to solve the problem of unbalanced data classification. For the existing random forest-based classification prediction model, its hyperparameters are dependent on empirical settings, which leads to the problem of unsatisfactory model performance. In order to make random forest find the optimum modelling corresponding to the character of unbalanced data sets and improve the accuracy of prediction, we apply the improved particle swarm optimization to set reasonable hyperparameters of the model. This paper proposes a random forest-based adaptive particle swarm optimization on data classification, and an adaptive particle swarm used to optimize the hyperparameters in the random forest to ensure that the model can better predict the unbalanced data accurately. Aiming at the premature convergence that appears in the particle swarm optimization algorithm, the population is adaptively divided according to the population fitness and the adaptive update strategy is introduced to enhance the ability of particles to jump out of the local optimum. Experimental results show that our proposed algorithms outperform the traditional ones, especially regarding the evaluation criterion of F1-measure and accuracy. The results on the six keel unbalanced data set the advantages of our proposed algorithms are presented.

Research on intelligent penetration mechanism of netgap data based on database adaptive matching

“…The investigation [11] found that population diversity and inertia weight are the keys to avoid premature convergence of PSO; the core guiding principles of the algorithm are clustering and adaptive adjustment of inertia weight. Through fast search clustering methods, the population can be adaptively divided into several subgroups to better explore the solution space of the problem.…”

Optimasi Metode Supervised Learning Dengan Menggunakan Particle Swarm Optimization Untuk Deteksi Malware

“…PSO merupakan algoritma Optimasi fitur dapat meningkatkan nilai akurasi yang dihasilkan PSO telah digunakan untuk optimasi peningkatan akurasi untuk Random forest, decision tree, naïve bayes dan KKN untuk klasifikasi dataset diabetes [15]. PSO dapat dikombinasikan dengan Teknik data sampling untuk meningkatkan akurasi dari algoritma an ensemble Classifier dalam klasifikasi [16]. Penelitian ini focus dan memberikan kotribusi keterbaruan diantaranya: membuat model klasifikasi kepuasaan layanan public menggunakan an ensemble classifier, menguji pengaruh pemilihan fitur terhadapat klasifikasi ensemble classifier, menguji pengaruh feature selection terhadap algoritma an ensemble classifier dan yang terakhir adalah kombinasi Teknik data sampling dan pemilihan fitur terhadap kinerja algoritma Ensemble classifier Dari latar belakang permasalahan dan studi literatur pada penelitian terdahulu, maka tujuan penelitian ini adalah untuk deteksi malware menggunakan algoritma supervised learning yaitu Random Forest (RF), XGBOOST, catboost dan lightgbm, yang sebelumnya akan dilakukan selection feature untuk meningkatkan nilai akurasi, sehingga model yang dihasilkan dapat direkomendasikan untuk deteksi malware.…”

Imbalanced Text Sentiment Classification Based on Multi-Channel BLTCN-BLSTM Self-Attention

“…For example, Irsoy et al [ 11 ] applied RNN for text sentiment orientation classification, Kim et al [ 12 ] used CNN for text sentiment orientation classification, and Soni et al [ 14 ] proposed TextConvoNet, a novel convolutional neural network (CNN)-based architecture for solving binary and multi-class text classification problems. At the same time, some scholars have begun to study the classification problem based on imbalanced data [ 15 , 16 ]. For example, Yin Hao et al [ 17 ] proposed a resampling multi-channel model, which randomly sampled the imbalanced samples to make the number of samples balanced before training.…”