A generic, fast and asymptotically efficient method for parametric estimation is described. It is based on the projected stochastic gradient descent on the log-likelihood function corrected by a single step of the Fisher scoring algorithm. We show theoretically and by simulations that it is an interesting alternative to the usual stochastic gradient descent with averaging or the adaptative stochastic gradient descent.

翻译：本文描述了一种通用、快速且渐近有效的参数估计方法。该方法基于对数似然函数的投影随机梯度下降，并通过Fisher得分算法的一步校正进行优化。我们从理论推导和仿真实验两方面证明，该方法相较于常规带平均的随机梯度下降法或自适应随机梯度下降法，是一种具有竞争力的替代方案。