Missingness Bias in Model Debugging
This addresses bias issues in model debugging tools for computer vision practitioners, but appears incremental as it adapts existing transformer methods to a specific problem.
The paper tackles the problem of bias introduced by heuristic methods like blacking out pixels when implementing missingness for model debugging in computer vision, and shows that transformer-based architectures enable a more natural implementation that improves reliability.
Missingness, or the absence of features from an input, is a concept fundamental to many model debugging tools. However, in computer vision, pixels cannot simply be removed from an image. One thus tends to resort to heuristics such as blacking out pixels, which may in turn introduce bias into the debugging process. We study such biases and, in particular, show how transformer-based architectures can enable a more natural implementation of missingness, which side-steps these issues and improves the reliability of model debugging in practice. Our code is available at https://github.com/madrylab/missingness