From 73fd6e6aafdb1b34e9a7349dca136f3ce3969ed4 Mon Sep 17 00:00:00 2001 From: Yuchen Pei Date: Mon, 3 Dec 2018 10:17:00 +0100 Subject: minor --- posts/2018-12-02-lime-shapley.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'posts/2018-12-02-lime-shapley.md') diff --git a/posts/2018-12-02-lime-shapley.md b/posts/2018-12-02-lime-shapley.md index 233de19..152036b 100644 --- a/posts/2018-12-02-lime-shapley.md +++ b/posts/2018-12-02-lime-shapley.md @@ -268,7 +268,7 @@ becomes $$v(S) = \mathbb E_{z_{N \setminus S} \sim \mu_{N \setminus S}} f(x_S, z_{N \setminus S}) \qquad (8)$$ -For example, Strumbelj-Kononenko (2010) considers this where $\mu$ is +For example, Strumbelj-Kononenko (2010) considers this scenario where $\mu$ is the uniform distribution over $X$, see Definition 4 there. A further simplification is model linearity, which means $f$ is linear. @@ -300,7 +300,7 @@ Evaluating SHAP --------------- The quest of the SHAP paper can be decoupled into two independent components: -the niceties of Shapley values and the choice of the coalitional game $v$. +showing the niceties of Shapley values and choosing the coalitional game $v$. The SHAP paper argues that Shapley values $\phi_i(v)$ are a good measurement because they are the only values satisfying the some nice properties including the Efficiency -- cgit v1.2.3