Influence Function: Local Robustness and Efficiency
We propose a direct approach to calculating influence functions based on the concept of functional derivatives. The relative simplicity of our direct method is demonstrated through well-known examples. Using influence functions as a key device, we examine the connection and difference between local robustness and efficiency in both joint and sequential identification/estimation procedures. We show that the joint procedure is associated with efficiency, while the sequential procedure is linked to local robustness. Furthermore, we provide conditions that are theoretically verifiable and empirically testable on when efficient and locally robust estimation for the parameter of interest in a semiparametric model can be achieved simultaneously. In addition, we present straightforward conditions for an adaptive procedure in the presence of nuisance parameters.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Efficient and Robust Estimation of the Generalized LATE Model
This paper studies the estimation of causal parameters in the generalized local average treatment effect (GLATE) model, a generalization of the classical LATE model encompassing multi-valued treatment and instrument. We …
Huber-Wasserstein barycenters for robust distribution-valued data
We propose a robust barycenter for distribution-valued data by incorporating the Huber loss directly into the optimal transport cost. In contrast to metric-space Huber means, which apply the Huber loss to the Wasserstein…
Directional Influence Function: Estimating Training Data Influence in Constrained Learning
As constrained learning becomes increasingly common, models are trained under explicit feasibility requirements to enforce fairness, safety, robustness, regulariza- tion, and physics or logic constraints. Understanding h…
Quantitative Robustness of Localized Support Vector Machines
The huge amount of available data nowadays is a challenge for kernel-based machine learning algorithms like SVMs with respect to runtime and storage capacities. Local approaches might help to relieve these issues and to …
Boosting in the presence of outliers: adaptive classification with non-convex loss functions
This paper examines the role and efficiency of the non-convex loss functions for binary classification problems. In particular, we investigate how to design a simple and effective boosting algorithm that is robust to the…
Binary ClassificationGeneral Classification