Related Experiment Videos
A Prediction Model Integrating Adaptive-Network-Based Fuzzy Inference System and Fuzzy C-Mean Clustering
IEEE Transactions on Cybernetics
|May 4, 2026
Summary
This study introduces ANFIS-CPM-FCM, a novel combined prediction model that improves multivariate prediction accuracy by integrating fuzzy clustering and adaptive networks. It overcomes limitations of existing models by considering predictor-output relationships for enhanced control system design.
Area of Science:
- Control Systems Engineering
- Artificial Intelligence
- Data Mining
Background:
- Multivariate prediction is vital for control systems, but high-dimensional data poses challenges.
- Existing combined prediction models (CPMs) often lose information via dimensionality reduction or ignore predictor-output relationships, limiting accuracy.
Purpose of the Study:
- To develop an advanced combined prediction model that enhances prediction accuracy and robustness.
- To address the limitations of existing CPMs by incorporating improved clustering and adaptive fuzzy inference.
Main Methods:
- Proposed an adaptive-network-based fuzzy inference system CPM (ANFIS-CPM) integrated with an improved fuzzy C-means (FCM) clustering algorithm (ANFIS-CPM-FCM).
- Developed a similarity metric for feature relationships and enhanced FCM for automatic optimal cluster determination using a density clustering model.
- Trained individual ANFIS models per cluster and aggregated predictions considering predictor-output sequence relationships.
Main Results:
- ANFIS-CPM-FCM demonstrated superior prediction accuracy and robustness compared to existing methods across six datasets.
- The integration of improved clustering with adaptive fuzzy inference systems proved beneficial for prediction tasks.
Conclusions:
- The proposed ANFIS-CPM-FCM effectively enhances multivariate prediction for control system design.
- The method offers a significant advancement over traditional CPMs by optimizing data handling and prediction aggregation.
