Accuracy Comparison of Enhanced Multi‐Layer Perceptron and Polynomial Regression in Food Calorie Measurement

This study explores the use of machine learning to improve the accuracy of dietary calorie estimation. It focuses on two approaches: the Novel Enhanced Multi-Layer Perceptron (MLP) and Polynomial Regression. These models were evaluated using different training and testing splits to predict calorie content. The experiments were conducted with a sample size optimized using GPower software, set at α=0.05 and power=0.85. The results demonstrated the superior performance of the Enhanced MLP, which achieved an accuracy of 97.5%. This was significantly higher than the 94.0% accuracy of Polynomial Regression. Statistical analysis using independent t-tests confirmed the significance of these results, with a p-value of 0.000. The Enhanced MLP's ability to model complex, non-linear relationships contributed to its improved accuracy and reliability. The study highlights the potential of advanced machine learning techniques in addressing challenges in calorie estimation. By leveraging robust architectures like the Enhanced MLP, it is possible to achieve greater precision and consistency. These findings have implications for various applications, including healthcare and nutrition monitoring. Future research should focus on optimizing training processes and expanding datasets to further enhance model performance.