Sample Efficient Model Evaluation
This addresses the practical bottleneck of labeling data for model evaluation, offering a more efficient method for researchers and practitioners in machine learning.
The paper tackles the problem of efficiently estimating test metrics like accuracy and F1 scores with minimal labeled data by comparing Importance Sampling and a novel application of Poisson Sampling, showing that Poisson Sampling outperforms Importance Sampling both theoretically and experimentally.
Labelling data is a major practical bottleneck in training and testing classifiers. Given a collection of unlabelled data points, we address how to select which subset to label to best estimate test metrics such as accuracy, $F_1$ score or micro/macro $F_1$. We consider two sampling based approaches, namely the well-known Importance Sampling and we introduce a novel application of Poisson Sampling. For both approaches we derive the minimal error sampling distributions and how to approximate and use them to form estimators and confidence intervals. We show that Poisson Sampling outperforms Importance Sampling both theoretically and experimentally.