论文标题
rao-blackwelling直通牙龈 - 宽符号梯度估计器
Rao-Blackwellizing the Straight-Through Gumbel-Softmax Gradient Estimator
论文作者
论文摘要
具有离散潜在变量的模型中的梯度估计是一个具有挑战性的问题，因为最简单的无偏估计器往往具有很高的差异。为了抵消这一点，现代估计器要么引入偏见，依靠多函数评估，要么使用所学的输入依赖性基准。因此，需要估计器需要最小的调整，计算便宜，并且平均误差较低。在本文中，我们表明，可以通过rao-blackwellization降低流行的gumbel-softmax估计量的直通变体的方差，而无需增加功能评估的数量。事实证明，这可以减少平方误差。我们从经验上证明，这会导致降低差异，更快的收敛性以及在两个无监督的潜在变量模型中的性能通常改善。
Gradient estimation in models with discrete latent variables is a challenging problem, because the simplest unbiased estimators tend to have high variance. To counteract this, modern estimators either introduce bias, rely on multiple function evaluations, or use learned, input-dependent baselines. Thus, there is a need for estimators that require minimal tuning, are computationally cheap, and have low mean squared error. In this paper, we show that the variance of the straight-through variant of the popular Gumbel-Softmax estimator can be reduced through Rao-Blackwellization without increasing the number of function evaluations. This provably reduces the mean squared error. We empirically demonstrate that this leads to variance reduction, faster convergence, and generally improved performance in two unsupervised latent variable models.
