Shapley value attribution is an increasingly popular explainable AI (XAI)
method, which quantifies the contribution of each feature to the model's
output. However, recent work has shown that most existing methods to implement
Shapley value attributions have some drawbacks. Due to these drawbacks, the
resulting Shapley value attributions may provide biased or unreliable
explanations, which fail to correctly capture the true intrinsic relationships
between features and model outputs. Moreover, it is difficult to evaluate these
explanation errors because the true underlying dependencies between features
and model outputs are typically unknown. In this paper, we theoretically
analyze the explanation errors of Shapley value attributions by decomposing the
explanation error into two components: observation bias and structural bias. We
also clarify the underlying causes of these two biases and demonstrate that
there is a trade-off between them. Based on this error analysis framework, we
develop two novel concepts: over-informative and under-informative
explanations. We theoretically analyze the potential over-informativeness and
under-informativeness of existing Shapley value attribution methods.
Particularly for the widely deployed assumption-based Shapley value
attributions, we affirm that they can easily be under-informative due to the
distribution drift caused by distributional assumptions. We also propose a
measurement tool to quantify the distribution drift that causes such errors.

本文分析了 Shapley 值归因的解释误差，将解释误差分解为观察偏差和结构偏差两个组成部分，并且证明它们之间存在权衡关系。基于此误差分析框架，提出了过多信息和过少信息解释这两个新概念，并对现有的 Shapley 值归因方法进行了可能的过多信息和过少信息的理论分析。

基于 Shapley 值的模型解释的错误分析：一种信息性视角

Error Analysis of Shapley Value-Based Model Explanations: An Informative  Perspective

We consider a variant of matrix completion where entries are revealed in a
biased manner, adopting a model akin to that introduced by Ma and Chen. Instead
of treating this observation bias as a disadvantage, as is typically the case,
our goal is to exploit the shared information between the bias and the outcome
of interest to improve predictions. Towards this, we propose a simple two-stage
algorithm: (i) interpreting the observation pattern as a fully observed noisy
matrix, we apply traditional matrix completion methods to the observation
pattern to estimate the distances between the latent factors; (ii) we apply
supervised learning on the recovered features to impute missing observations.
We establish finite-sample error rates that are competitive with the
corresponding supervised learning parametric rates, suggesting that our
learning performance is comparable to having access to the unobserved
covariates. Empirical evaluation using a real-world dataset reflects similar
performance gains, with our algorithm's estimates having 30x smaller mean
squared error compared to traditional matrix completion methods.

在 Ma 和 Chen 引入的模型的基础上，本文提出了一种两阶段的算法来处理矩阵填充中的观测偏差，并利用共享信息来提高预测性能，经实验证明本算法的表现可与未观测协变量相同，并获得性能提高。

利用观测偏差改进矩阵补全

Exploiting Observation Bias to Improve Matrix Completion

We propose using recognition networks for approximate inference inBayesian
networks (BNs). A recognition network is a multilayerperception (MLP) trained
to predict posterior marginals given observedevidence in a particular BN. The
input to the MLP is a vector of thestates of the evidential nodes. The activity
of an output unit isinterpreted as a prediction of the posterior marginal of
thecorresponding variable. The MLP is trained using samples generated fromthe
corresponding BN.We evaluate a recognition network that was trained to do
inference ina large Bayesian network, similar in structure and complexity to
theQuick Medical Reference, Decision Theoretic (QMR-DT). Our networkis a
binary, two-layer, noisy-OR network containing over 4000 potentially observable
nodes and over 600 unobservable, hidden nodes. Inreal medical diagnosis, most
observables are unavailable, and there isa complex and unknown bias that
selects which ones are provided. Weincorporate a very basic type of selection
bias in our network: a knownpreference that available observables are positive
rather than negative.Even this simple bias has a significant effect on the
posterior. We compare the performance of our recognition network
tostate-of-the-art approximate inference algorithms on a large set oftest
cases. In order to evaluate the effect of our simplistic modelof the selection
bias, we evaluate algorithms using a variety ofincorrectly modeled observation
biases. Recognition networks performwell using both correct and incorrect
observation biases.

使用识别网络进行贝叶斯网络的近似推理，以实现媲美最先进的近似推理算法，以解决在医学诊断中存在的各种观测偏差。