The Adoption of Deep Belief Network Classifier with Shark Smell Optimizer to Predict the Soil Liquefaction

In this paper, post-liquefaction Standard penetration test (SPT) data was collected from the Chi-Chi earthquake and Deep Belief Network (DBN) model using a metaheuristic-based shark smell optimizer (SSO) to improve the predicted accuracy by many iterations. First, the data were normalized and the correlation between variables was determined using chi-square. Next, random sampling was used to select the parameters that were used to train and test the DBN models. Lastly, the SSO was implemented to improve the model’s accuracy; it was found that SSO not only improved the model’s fitting/accuracy but also sped up the process.