Stochastic gradient descent (SGD) provides a simple and efficient way to solve a broad range of machine learning problems. Here, we focus on distribution regression (DR), involving two stages of sampling : Firstly, we regress from probability measures to real-valued responses. Secondly, we sample bags from these distributions for utilizing them to solve the overall regression problem.
Recently, DR has been tackled by applying kernel regularized least-squares and the learning properties of this approach are well understood. However, nothing is known about the learning properties of SGD for two stage sampling problems in an RKHS framework. We fill this gap and provide theoretical guarantees for the performance of SGD for DR. Our bounds are optimal in a mini-max sense under standard assumptions, see [Müe21].

muecke@math.tu-berlin.de

[Müe21] Nicole Müecke. Stochastic gradient descent meets distribution regression. In The 24th International Conference on Artificial Intelligence and Statistics. PMLR, 2021.