Figure 5  The average scores with the corresponding first tier and second tier improvements on synthetic data using the random starts with Exit-point approach with different (l, d) motifs.