Statistical Significance Results for MaxGRNet Model Evaluation

Md Mehedi Hasan Santo published a statistical significance summary comparing the MaxGRNet model to baselines on dataset DS1. The summary includes results from 5-fold paired tests with Holm correction for prediction and per-image paired tests for explainability metrics. The dataset was uploaded to figshare on April 8, 2026.

Use Cases

• Compare Insertion AUC scores between MaxGRNet and GaussianBlur baseline models on common test images.
• Analyze 1–Deletion AUC results from per-image paired tests under the correct-only protocol.
• Evaluate statistical significance of prediction performance differences using 5-fold paired test results.
• Assess the impact of Holm correction on the significance of explainability metric comparisons.

Strengths

• Statistical tests include a rigorous 5-fold paired design with Holm correction.
• Explainability evaluation uses two specific metrics: Insertion AUC and 1–Deletion AUC.
• Data is openly licensed under CC-BY-4.0.

Limitations

• Dataset size is only 5.5 KB, indicating a very limited scope of results.
• No column names or sample data are provided, limiting detailed analysis.
• The underlying dataset DS1 and the specific baseline models are not described.

Provenance

Source

Md Mehedi Hasan Santo via figshare.

Collection Method

Results from statistical tests comparing a machine learning model (MaxGRNet) to baselines.

Freshness

Last updated in April 2026.