Müntz-Szász Networks (MSN) are a novel neural network architecture that replaces fixed activation functions with learnable fractional power bases. Unlike standard MLPs that use fixed activations (ReLU ...