Fig. 3

The first set of three figures, labeled as (b), illustrates the variables (u, v, p) for the neural network post-variance reduction. In contrast, figures labeled as (a) represent the neural network initialized using the Xavier scheme. The final set of three figures exhibits the value of du/dx for two scenarios: one for the neural network after variance reduction (d), and the other for a neural network initialized using the Xavier scheme(c). Please note that all the figures presented are post-variance reduction and prior to the commencement of any training.