A Rigorous Behavior Assessment of CNNs Using a Data-Domain Sampling Regime
This provides insights into CNN performance for visualization interpretation, but it is incremental as it focuses on a specific domain.
The study tackled the problem of quantifying CNNs' behavior in interpreting bar charts, finding that CNNs can outperform humans and their biases depend on training-test distribution distance, based on analysis of 16 million trials from 800 CNN models and 6,825 human trials.
We present a data-domain sampling regime for quantifying CNNs' graphic perception behaviors. This regime lets us evaluate CNNs' ratio estimation ability in bar charts from three perspectives: sensitivity to training-test distribution discrepancies, stability to limited samples, and relative expertise to human observers. After analyzing 16 million trials from 800 CNNs models and 6,825 trials from 113 human participants, we arrived at a simple and actionable conclusion: CNNs can outperform humans and their biases simply depend on the training-test distance. We show evidence of this simple, elegant behavior of the machines when they interpret visualization images. osf.io/gfqc3 provides registration, the code for our sampling regime, and experimental results.