Unsupervised feature selection via self-paced learning and low-redundant regularization

Optimized Machine Learning Model for Predicting Compressive Strength of Alkali-Activated Concrete Through Multi-Faceted Comparative Analysis

Alkali-activated concrete (AAC), produced from industrial by-products like fly ash and slag, offers a promising alternative to traditional Portland cement concrete by significantly reducing carbon emissions. Yet, the inherent variability in AAC formulations presents a challenge for accurately predicting its compressive strength using conventional approaches. To address this, we leverage machine learning (ML) techniques, which enable more precise strength predictions based on a combination of material properties and cement mix design parameters. In this study, we curated an extensive dataset comprising 1756 unique AAC mixtures to support robust ML-based modeling. Four distinct input variable schemes were devised to identify the optimal predictor set, and a comparative analysis was performed to evaluate their effectiveness. After this, we investigated the performance of several popular ML algorithms, including random forest (RF), adaptive boosting (AdaBoost), gradient boosting regression trees (GBRTs), and extreme gradient boosting (XGBoost). Among these, the XGBoost model consistently outperformed its counterparts. To further enhance the predictive accuracy of the XGBoost model, we applied four state-of-the-art optimization techniques: the Gray Wolf Optimizer (GWO), Whale Optimization Algorithm (WOA), beetle antennae search (BAS), and Bayesian optimization (BO). The optimized XGBoost model delivered superior performance, achieving a remarkable coefficient of determination (R2) of 0.99 on the training set and 0.94 across the entire dataset. Finally, we employed SHapely Additive exPlanations (SHAP) to imbue the optimized model with interpretability, enabling deeper insights into the complex relationships governing AAC formulations. Through the lens of ML, we highlight the benefits of the multi-faceted synergistic approach for AAC strength prediction, which combines careful input parameter selection, optimal hyperparameter tuning, and enhanced model interpretability. This integrated strategy improves both the robustness and scalability of the model, offering a clear and reliable prediction of AAC performance.

Self-paced regularized adaptive multi-view unsupervised feature selection

Multi-view unsupervised feature selection (MUFS) is an efficient approach for dimensional reduction of heterogeneous data. However, existing MUFS approaches mostly assign the samples the same weight, thus the diversity of samples is not utilized efficiently. Additionally, due to the presence of various regularizations, the resulting MUFS problems are often non-convex, making it difficult to find the optimal solutions. To address this issue, a novel MUFS method named Self-paced Regularized Adaptive Multi-view Unsupervised Feature Selection (SPAMUFS) is proposed. Specifically, the proposed approach firstly trains the MUFS model with simple samples, and gradually learns complex samples by using self-paced regularizer. l2,p-norm (0 Collapse

Key Words

• Hypergraph
• Multi-view unsupervised feature selection
• Self-paced learning
• Sparse learning
• l(2,p)-norm

Collapse

MESH Headings

• Unsupervised Machine Learning
• Algorithms
• Humans
• Neural Networks, Computer

Collapse

Grants Collapse

Affiliation(s)

Xuanhao Yang College of Electronic and Information Engineering, Southwest University, Chongqing, 400715, China.
Hangjun Che College of Electronic and Information Engineering, Southwest University, Chongqing, 400715, China; Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing, Chongqing, 400715, China.
Man-Fai Leung School of Computing and Information Science, Faculty of Science and Engineering, Anglia Ruskin University, Cambridge, UK.
Shiping Wen Faculty of Engineering and Information Technology, Australian Artificial Intelligence Institute, University of Technology Sydney, Sydney, NSW 2007, Australia.

Collapse

Collaborators Collapse

A constraint score guided meta-heuristic searching to attribute reduction

Essentially, the problem solving of attribute reduction can be regarded as a process of reduct searching which will be terminated if a pre-defined restriction is achieved. Presently, among a variety of searching strategies, meta-heuristic searching has been widely accepted. Nevertheless, it should be emphasized that the iterative procedures in most meta-heuristic algorithms rely heavily on the random generation of initial population, such a type of generation is naturally associated with the limitations of inferior stability and performance. Therefore, a constraint score guidance is proposed before carrying out meta-heuristic searching and then a novel framework to seek out reduct is developed. Firstly, for each attribute and each label in data, the index called local constraint score is calculated. Secondly, the qualified attributes are identified by those constraint scores, which consist of the foundation of initial population. Finally, the meta-heuristic searching can be further employed to achieve the required restriction in attribute reduction. Note that most existing meta-heuristic searchings and popular measures (evaluate the significance of attributes) can be embedded into our framework. Comprehensive experiments over 20 public datasets clearly validated the effectiveness of our framework: it is beneficial to reduct with superior stabilities, and the derived reduct may further contribute to the improvement of classification performance.

Bearing Fault Diagnosis Method Based on RCMFDE-SPLR and Ocean Predator Algorithm Optimizing Support Vector Machine

For the problem that rolling bearing fault characteristics are difficult to extract accurately and the fault diagnosis accuracy is not high, an unsupervised characteristic selection method of refined composite multiscale fluctuation-based dispersion entropy (RCMFDE) combined with self-paced learning and low-redundant regularization (SPLR) is proposed, for which the fault diagnosis is carried out by support vector machine (SVM) optimized by the marine predator algorithm (MPA). First, we extract the entropy characteristics of the bearings under different fault states by RCMFDE and the introduction of the fine composite multiscale coarse-grained method and fluctuation strategy improves the stability and estimation accuracy of the bearing characteristics; then, a novel dimensionality-reduction method, SPLR, is used to select better entropy characteristics, and the local flow structure of the fault characteristics is preserved and the redundancy is constrained by two regularization terms; finally, using the MPA-optimized SVM classifier by combining Levy motion and Eddy motion strategies, the preferred RCMFDE is fed into the MPA-SVM model for fault diagnosis, for which the obtained bearing fault diagnosis accuracy is 97.67%. The results show that the RCMFDE can effectively improve the stability and accuracy of the bearing characteristics, the SPLR-based low-dimensional characteristics can suppress the redundancy characteristics and improve the effectiveness of the characteristics, and the MPA-based adaptive SVM model solves the parameter randomness problem and, therefore, the proposed method has outstanding superiority.