“…For example, from a perspective of states, eigenstates of different qubits can be orthogonal; from a perspective of phase, on the Bloch sphere, the rotation operators about the x, ŷ and ẑ axes, i.e., R x (𝜃 x ), R y (𝜃 y ), and R z (𝜃 z ) are simply a rotation of the sphere about the corresponding axis by 𝜃 x , 𝜃 y , and 𝜃 z respectively, and the planes of these angles, which are parameters of quantum neural networks, are orthogonal to each other no matter what the angles are. [17] Based on the points discussed above, we assume that the characteristic of quantum states and orthogonality relations in quantum computing may partially explain why the generalizability of our quantum hybrid models has been improved compared to that of the classical ones.…”