Self-supervised Contrastive Learning for Chaotic Time-Series Classification

It is often crucial to compute bifurcation diagrams in order to assess the stability and effects of multiple parameters on the overall dynamical properties of nonlinear systems. The case of one-parameter bifurcation diagrams is fairly easy to deal with, but it is more difficult (due to computational requirements) to do so when two or more parameters change at once, as we need fairly advanced hardware and software resources for that purpose [1–4]