Data Models for Dataset Drift Controls in Machine Learning With Images
Abstract: Camera images are ubiquitous in machine learning research. They also play a central role in the delivery of important services spanning medicine and environmental surveying. However, the application of machine learning models in these domains has been limited because of robustness concerns. A primary failure mode are performance drops due to differences between the training and deployment data. While there are methods to prospectively validate the robustness of machine learning models to such dataset drifts, existing approaches do not account for explicit models of the primary object of interest: the data. This makes it difficult to create physically faithful drift test cases or to provide precise specifications of data models that should be avoided during the deployment of a machine learning model. In this study, we demonstrate how these shortcomings can be overcome by pairing machine learning robustness validation with physical optics. We examine the role raw sensor data and differentiable data models can play in controlling performance risks related to image dataset drift. The findings are distilled into three applications. First, drift synthesis enables the controlled generation of physically faithful drift test cases. The experiments presented here show that the average decrease in model performance is ten to four times less severe than under post-hoc augmentation testing. Second, the gradient connection between machine learning model and our data models allows for drift forensics that can be used to specify performance-sensitive data models which should be avoided during deployment of a machine learning model. Third, drift adjustment opens up the possibility for processing adjustments in the face of drift. This can lead to speed up and stabilization of classifier training at a margin of up to 20% in validation accuracy. Alongside our data model code we release two datasets to the public that we collected as part of this work. In total, the two datasets, Raw-Microscopy and Raw-Drone, comprise 1,488 scientifically calibrated reference raw sensor measurements, 8,928 raw intensity variations as well as 17,856 images processed through our data models with twelve different configurations. A guide to access the open code and datasets is available at https://anonymous.4open.science/r/tmlr/README.md.
Submission Length: Long submission (more than 12 pages of main content)
Assigned Action Editor: ~Behnam_Neyshabur1
Submission Number: 193
Loading