Table 11 Results from the utilization of raw and minimum–maximum preprocessing for 28-day compressive strength by using DAVO algorithm.

7-day compressive strength	Neuron Number	4	8	12	16
MSE test	Raw	223.55	201.31	194.99	225.06
MSE test	Min–Max	223.55	201.31	187.44	225.06
MSE train	Raw	224.04	208.77	210.24	226.48
MSE train	Min–Max	224.04	208.77	210.93	226.48
R test	Raw	0.61	0.66	0.68	0.60
R test	Min–Max	0.61	0.66	0.68	0.60
R train	Raw	0.56	0.60	0.60	0.55
R train	Min–Max	0.56	0.60	0.60	0.55

Note that, in this study, data normalization was employed to scale the input parameters to a common range, facilitating improved convergence and stability during the training of the DAVOA-ENN models. We used the Min–Max normalization technique, which transforms each input variable to a specific range, typically between 0 and 1. Min–Max normalization ensures that all input features contribute equally to the model training process by rescaling them to the same range. By transforming the data to a uniform scale, the optimization algorithm can achieve faster convergence and more stable solutions, reducing the risk of numerical instability during training. Further, this method maintains the relative relationships between data points, which is crucial for preserving the inherent patterns in the data.

Quick links

Search