Six Sigma For Neural Networks: Taguchi-based optimization
Sai Varun Kodathala
TL;DR
This work adapts Taguchi Design of Experiments to CNN hyperparameter optimization for professional boxing action recognition, employing an $L_{12}(2^{11})$ orthogonal array to evaluate eight binary factors across twelve trials. It introduces five multi-objective SNR formulations, including a novel logarithmic integration, to balance accuracy and loss metrics and identify robust configurations with substantially fewer experiments than exhaustive grid searches (approximately 95% fewer trials). The results show that a logarithmic multi-objective approach (Approach 3) offers the best trade-off, achieving high training accuracy ($98.84\%$) and solid validation accuracy ($86.25\%$) with low losses ($TL=0.0442$, $VL=0.5784$) and confirming learning rate as the most influential hyperparameter. The study demonstrates practical, interpretable, and computationally efficient optimization for CNNs in sports analytics, with potential generalization to broader neural network tuning tasks and other domains.
Abstract
The optimization of hyperparameters in convolutional neural networks (CNNs) remains a challenging and computationally expensive process, often requiring extensive trial-and-error approaches or exhaustive grid searches. This study introduces the application of Taguchi Design of Experiments methodology, a statistical optimization technique traditionally used in quality engineering, to systematically optimize CNN hyperparameters for professional boxing action recognition. Using an L12(211) orthogonal array, eight hyperparameters including image size, color mode, activation function, learning rate, rescaling, shuffling, vertical flip, and horizontal flip were systematically evaluated across twelve experimental configurations. To address the multi-objective nature of machine learning optimization, five different approaches were developed to simultaneously optimize training accuracy, validation accuracy, training loss, and validation loss using Signal-to-Noise ratio analysis. The study employed a novel logarithmic scaling technique to unify conflicting metrics and enable comprehensive multi-quality assessment within the Taguchi framework. Results demonstrate that Approach 3, combining weighted accuracy metrics with logarithmically transformed loss functions, achieved optimal performance with 98.84% training accuracy and 86.25% validation accuracy while maintaining minimal loss values. The Taguchi analysis revealed that learning rate emerged as the most influential parameter, followed by image size and activation function, providing clear guidance for hyperparameter prioritization in CNN optimization.