Chapter 19 Gradient Estimation

107 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Concrete Dropout (2017, background): “Several estimators are available for us to do this: for example the score function estimator (also known as a likelihood ratio estimator and Reinforce [6, 12, 30, 35]), or the pathwise derivative estimator (this estimator is also referred to in the literature as the re-parametrisation trick, infinit”
A high-bias, low-variance introduction to Machine Learning for physicists (2018, unknown): “…model, we can also easily calculate the gradient of any expectation value 〈f(x)〉model using what is commonly called the “log-derivative trick” in ML (Fu, 2006; Kleijnen and Rubinstein, 1996): ∂ ∂θi 〈f(x)〉model = ∫ dx ∂pθ(x) ∂θi f(x) = 〈∂ log pθ(x) ∂θi f(x) 〉 model = 〈Oi(x)f(x)〉model ≈ ∑…”
A Guide to Sample Average Approximation (2015, background): “If we take as given that SAA-appropriate problems are those for which both the approximating functions fn(·) and f(·) are differentiable, and the derivatives of fn(·) converge to those for f(·) then we arrive at an underlying theme of this section, which is the following “meta principle:” SAA-approp”