Skip to main content
. 2023 Aug 21;15(1):2244214. doi: 10.1080/19420862.2023.2244214

Figure 5.

Scatterplot A is showing visualization of embedding sequences space using PCA. MTS gradually explores from around the training sequences. S explores a more diverse sequence space. Boxplot B is showing Euclid distance in embedding space from the nearest sequence in the training data at each round. MTS selects candidates in the vicinity of the training sequences and TS picks them far from the training sequences.

Characteristics of sequence space explored during Bayesian optimization. (A) Visualization of embedding sequence space via PCA. Green points represent initial training data. Redish points are the sequences of MTS. Bluish points are the sequences of TS. (B) Comparison of Euclid distance in embedding space from the nearest sequence in the training data at each round. Redish boxplot represent the distribution of Euclid distance at each round on MTS. Bluish boxplot represent the distribution of Euclid distance at each round on TS.