Figure 5
From: Reduced order modeling for flow and transport problems with Barlow Twins self-supervised learning
Example 4 results: the moving average (a window size of 50) of mean squared error (MSE) of (a) BT–AE 4 Q, (b) BT–AE 16 Q, (c) BT–AE 256 Q (please refer to Table 2). (d) Data compression loss for validation set (Eq. 18), (e) Barlow Twins loss for validation set (Eq. 14), (f) mapping using ANN loss for validation set (Eq. 19), and latent space plot of BT–AE 16 Q model for (g) \(Ra_1\) and (h) \(Ra_4\). Latent space plots are constructed using t-Distributed Stochastic Neighbor Embedding (t-SNE). Different colors represent each value of \(\textrm{Ra}\) value. We calculate the t-SNE plots using Scikit-Learn package using its default setting and perplexity of 15.
