Effect of different hyperparameters using non-Gaussian synthetic data - Lognormal

Effects of different parameters using 2D non-Gaussian (log-normal, μ=0, and σ=1) data with 10K observations. Note that the effects of the parameters are similar to those of Figure 1. However, LGC with too large values of T and α is prone to outliers (long tails), as shown in (d) and (l), but this problem can be solved by setting a larger value of k.