Model Transferability with Responsive Decision Subjects
Model Transferability with Responsive Decision Subjects
复制标题
DOI：
--
复制
发表时间：
2021-07
期刊：
影响因子：
--
通讯作者：
Yang Liu;Yatong Chen;Zeyu Tang;Kun Zhang
中科院分区：
文献类型：
--
作者：
Yang Liu;Yatong Chen;Zeyu Tang;Kun Zhang
给定一个算法预测器，它对由人类战略决策主体组成的某些源群体是准确的，如果群体对其做出反应，它还会保持准确吗？在我们的设置中，代理或用户对应于从分布 $\cal{D}$ 中抽取的样本 $(X,Y)$，并将面对模型 $h$ 及其分类结果 $h(X)$。代理可以修改$X$以适应$h$，这将导致$(X,Y)$上的分布变化。我们的制定是受到应用程序的启发，其中部署的机器学习模型受到人类代理的影响，并且最终将面临响应式和交互式数据分布。我们通过研究在可用源分布（数据）上训练的模型的性能如何转化为其诱导域上的性能，来正式讨论模型的可转移性。我们提供了由于诱导域转移而导致的性能差距的上限，以及分类器在源训练分布或诱导目标分布上必须承受的权衡的下限。我们为两种流行的域适应设置提供了进一步的实例化分析，包括协变量偏移和目标偏移。
Given an algorithmic predictor that is accurate on some source population consisting of strategic human decision subjects, will it remain accurate if the population respond to it? In our setting, an agent or a user corresponds to a sample $(X,Y)$ drawn from a distribution $\cal{D}$ and will face a model $h$ and its classification result $h(X)$. Agents can modify $X$ to adapt to $h$, which will incur a distribution shift on $(X,Y)$. Our formulation is motivated by applications where the deployed machine learning models are subjected to human agents, and will ultimately face responsive and interactive data distributions. We formalize the discussions of the transferability of a model by studying how the performance of the model trained on the available source distribution (data) would translate to the performance on its induced domain. We provide both upper bounds for the performance gap due to the induced domain shift, as well as lower bounds for the trade-offs that a classifier has to suffer on either the source training distribution or the induced target distribution. We provide further instantiated analysis for two popular domain adaptation settings, including covariate shift and target shift.