Taguchi Based Design

Taguchi-based design is a robust optimization methodology used to efficiently determine optimal parameters for complex systems, particularly in machine learning applications. Current research focuses on applying this method to tune hyperparameters and architectures of convolutional neural networks (CNNs), improving their performance in diverse tasks like image classification and material property prediction. This approach enhances the efficiency and robustness of model development, leading to improved accuracy and reduced computational costs across various engineering and scientific domains. The resulting optimized models offer significant advantages in applications ranging from automated defect detection to material science analysis.