Abstract | ||
---|---|---|
In many applications we seek to maximize an expectation with respect to a distribution over discrete variables. Estimating gradients of such objectives with respect to the distribution parameters is a challenging problem. We analyze existing solutions including finite-difference (FD) estimators and continuous relaxation (CR) estimators in terms of bias and variance. We show that the commonly used Gumbel-Softmax estimator is biased and propose a simple method to reduce it. We also derive a simpler piece-wise linear continuous relaxation that also possesses reduced bias. We demonstrate empirically that reduced bias leads to a better performance in variational inference and on binary optimization tasks. |
Year | Venue | Field |
---|---|---|
2018 | arXiv: Machine Learning | Mathematical optimization,Inference,Binary optimization,Mathematics,Estimator |
DocType | Volume | Citations |
Journal | abs/1810.00116 | 1 |
PageRank | References | Authors |
0.35 | 0 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Evgeny Andriyash | 1 | 12 | 2.80 |
Vahdat, Arash | 2 | 353 | 18.20 |
William G. Macready | 3 | 161 | 39.07 |