Figure 1
From: Experimental kernel-based quantum machine learning in finite feature space
Kernel family (2) for different amplitude choices. (a) We find that the resolution-optimized kernel (blue solid) exhibits suppressed side maxima as compared to the MSI kernel (red dashed), while the TSQ kernel (with squeezing factor \(\zeta =2\), black dotted) maintains a nonvanishing plateau at all x values. For comparison, we also display the respective squeezed-state kernel for \(N \rightarrow \infty\) (gray dotted) and CK (purple dash-dotted). (b) Characteristic amplitude progressions for the example of \(N=14\) and \(\zeta =4\). (c) The optimized kernel exhibits a significantly improved resolution progression with N, as compared to the MSI or the TSQ kernel (here with \(\zeta =3\)).
