Does the dataset meet your expectations? Explaining sample representation in image data
This work addresses the problem of explaining sample representation deficiencies in image datasets for machine learning practitioners, particularly when data labeling is expensive.
This paper introduces a method to identify and explain deficiencies in training data diversity by comparing the actual distribution of annotations in a dataset with a manually specified expected distribution. It leverages parametric simulation to map the expected annotation distribution into test samples, explaining representation mismatches between simulated and collected data. The method was applied to a dataset of geometric shapes, providing qualitative and quantitative explanations of sample representation based on attributes like size, position, and pixel brightness.
Since the behavior of a neural network model is adversely affected by a lack of diversity in training data, we present a method that identifies and explains such deficiencies. When a dataset is labeled, we note that annotations alone are capable of providing a human interpretable summary of sample diversity. This allows explaining any lack of diversity as the mismatch found when comparing the \textit{actual} distribution of annotations in the dataset with an \textit{expected} distribution of annotations, specified manually to capture essential label diversity. While, in many practical cases, labeling (samples $\rightarrow$ annotations) is expensive, its inverse, simulation (annotations $\rightarrow$ samples) can be cheaper. By mapping the expected distribution of annotations into test samples using parametric simulation, we present a method that explains sample representation using the mismatch in diversity between simulated and collected data. We then apply the method to examine a dataset of geometric shapes to qualitatively and quantitatively explain sample representation in terms of comprehensible aspects such as size, position, and pixel brightness.