.png){kind=icon}
.png){kind=icon}
Return
Optimized machine learning approach for structural response prediction using wolf-bird optimizer
.png){kind=icon}
OA
AI
DOI：10.1016/j.istruc.2024.106691.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
One of the most well-known hybrid machine learning approaches is the Adaptive Neuro Fuzzy Inference System (ANFIS), combining neural networks and fuzzy logic, which creates a system for learning and decision making based on the provided data. Despite ANFIS's outstanding capability in handling uncertainty and imprecision in provided data set, better interpreting complex systems, and integrating human expertise through linguistic rules, there are some major challenges in using this method in different applications. Sensitivity to initial parameters, potential overfitting or under fitting issues, and difficulty in effectively capturing highly complex nonlinear interactions in certain applications makes this method's applicability very challenging in dealing with complex data sets. One of the frequent ways in enhancing the ANFIS's performance is to use metaheuristic algorithms for optimal parameter tuning of this system. For this purpose, the recently proposed Wolf-Bird Optimizer (WBO) is utilized alongside the well-known approaches form the literature to develop a prediction problem. The 9-story benchmark building structure is utilized as the design example, while the optimized ANFIS schemes are utilized for predicting the structural response by using different input data sets. Based on the results, ANFIS doesn't provide better prediction by using diverse data sets so that the use of metaheuristics can enhance its overall performance. The WBO provides better prediction than other metaheuristic approaches.
Keywords:
Machine learning
Metaheuristic algorithms
Adaptive Neuro Fuzzy Inference System
Optimization
Structural response prediction
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.