研究目的
Investigating the performance of a regularized weighted circular complex-valued extreme learning machine (RWCC-ELM) for imbalanced learning.
研究成果
RWCC-ELM outperforms CC-ELM and WELM for most evaluated datasets, demonstrating its effectiveness in handling imbalanced learning problems. Future work may include applying RWCC-ELM to real-world applications with complex-valued inputs.
研究不足
The study is limited to datasets from the Keel repository and does not explore the application of RWCC-ELM on real-world complex-valued input datasets with large class distribution variations.
1：Experimental Design and Method Selection
The study proposes RWCC-ELM, which combines the strengths of CC-ELM and WELM. It uses a circular transformation function to map real-valued data to complex domain and employs a fully complex sech activation function in the hidden layer.
2：Sample Selection and Data Sources
The performance of RWCC-ELM is evaluated using imbalanced datasets from the Keel repository, including 20 binary and 10 multiclass datasets with varying imbalance ratios.
3：List of Experimental Equipment and Materials
Matlab 7.1 running on PC with Intel core i5 processor, 3.20 GHz CPU and 2 GB RAM.
4：Experimental Procedures and Operational Workflow
The study involves mapping real-valued data to complex domain, computing the target matrix, choosing the number of neurons and regularization parameter, initializing weights, computing the hidden layer output matrix, and calculating weights between hidden and output layers.
5：Data Analysis Methods
Performance is evaluated using G-mean and overall accuracy. Wilcoxon signed-rank test is conducted for statistical comparison.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
