When Does Interference Help Learning? Kernel Geometry as a Pre-Experimental Test for Photonic Reservoir Computing
The paper investigates whether multi-photon interference is a useful resource for photonic quantum machine learning by analyzing kernel geometry. It found that interference creates a large geometric separation between feature maps, suggesting an interference-specific learning resource exists. However, this advantage is only accessible if the task aligns with this geometry, meaning task alignment determines whether the quantum method provides a performance boost.