《Understanding Black-box Predictions via Influence Functions》笔记
[1]通过升权(§2.1\S2.1§2.1)、扰动(§2.2\S2.2§2.2)两种方式研究某个trainingpointz=(x,y)z=(x,y)z=(x,y)对(优化后的)模型参数θ^\hat\thetaθ^、模型在某个testpointztest=(xtest,ytest)z_\text{test}=(x_\text{test},y_\text{test})ztest=(xtest,yte