Machine learning detection of Majorana zero modes from zero-bias peak measurements

Majorana zero modes (MZMs) carry non-Abelian statistics and great promise for topological quantum computation. A key signature of MZMs is the zero-bias peaks (ZBPs) in tunneling differential conductance. However, identifying MZMs from ZBPs has been challenging due to topological trivial states generating spurious signals. In this work, we introduce a machine learning framework that can distinguish MZM from other signals using ZBP data. Quantum transport simulation from tight-binding models is used to generate training data, while persistent cohomology analysis confirms the feasibility of machine-based classification. Even with noisy data, the extreme gradient boosting (XGBoost) classifier reaches 85 % accuracy for 1D data and 94 % for 2D data with Zeeman splitting. Tests on prior experiments show that key observations from some of the prior experiments are more likely to originate from MZMs. Our model offers a quantitative approach to assess MZMs using solely ZBP data. Furthermore, our results highlight the use of machine learning on exotic quantum systems with experimental-computational integration.