interpretabilityfactbullishOperator learning can be reformulated as a linear combination of generalized functional linear models to achieve direct interpretabilityMachine Learning27 Jul 2026http://arxiv.org/abs/2607.02203v1