Enforcing exact permutation and rotational symmetries in the application of quantum neural networks on point cloud datasets

Recent developments in the field of quantum machine learning have promoted the idea of incorporating physical symmetries in the structure of quantum circuits. A crucial milestone in this area is the realization of <a:math xmlns:a="http://www.w3.org/1998/Math/MathML"><a:msub><a:mi>S</a:mi><a:mi>n</a:mi></a:msub></a:math>-permutation equivariant quantum neural networks (QNNs) that are equivariant under permutations of input objects. In this paper, we focus on encoding the rotational symmetry of point cloud datasets into the QNN. The key insight of the approach is that all rotationally invariant functions with vector inputs are equivalent to a function with inputs of vector inner products. We provide a structure of the QNN that is exactly invariant to both rotations and permutations, with its efficacy demonstrated numerically in the problems of two-dimensional image classifications and identifying high-energy particle decays, produced by proton-proton collisions, with the <b:math xmlns:b="http://www.w3.org/1998/Math/MathML"><b:mrow><b:mi>SO</b:mi><b:mo>(</b:mo><b:mn>1</b:mn><b:mo>,</b:mo><b:mn>3</b:mn><b:mo>)</b:mo></b:mrow></b:math> Lorentz symmetry. Published by the American Physical Society 2024