Comparative Analysis of Quantum Feature Selection Methods: QUBOFS vs. VarQFS

In machine learning and data analysis, features selection is one important step to reduce the complexity and improve the model performance. However, finding the optimal feature subset for a given task is NPhard, due its huge underlying solution space. This makes exact feature selection on classical hardware usually infeasible and rely on approximations. In this poster, we compare two Quantum feature selection strategies, one based on the QUBO formulation [1] and one based on blackbox optimization [2] on two data sets. A logistic regression classifier is used to evaluate the selected features.