State-of-art minibatches via novel DPP kernels: discretization, wavelets, and rough objectives

ArXi:2605.13127v1 Announce Type: cross Determinantal point processes (DPPs) have emerged as a kernelized alternative to vanilla independent sampling for generating efficient minibatches, coresets and other parsimonious representations of large-scale datasets. While theoretical foundations and promising empirical performance have been nstrated, there are two challenges for current proposals for DPP-based coresets or minibatches.