Prediction of Ultimate Tensile Strength of Additive Manufactured Specimens using Neurosymbolic based Machine Learning Algorithm.

In this research, a novel Neurosymbolic approach that seamlessly integrates the power of neural networks with the interpretability of symbolic AI to tackle complex problem domains. This pioneering algorithm not only outperforms traditional Artificial Neural Network (ANN) models but also offers enhanced insights and understanding of the underlying data patterns. The comprehensive evaluation of our Neurosymbolic approach reveals lower Mean Squared Error (MSE) values and higher R-squared (R2) scores, signifying its superior ability to predict outcomes and explain the variance in output values. Notably, the model exhibits minimal discrepancies between training and validation performance, indicating its robust generalization capabilities to new, unseen data. The Neurosymbolic model achieves an R2 score of 0.9871 on the training dataset and 0.9762 on the validation dataset, while the Simple ANN model yields an R2 score of 0.9780 on the training dataset and 0.9759 on the validation dataset.