hypertower/v4/figures/output/S8e_clinical_importance.csv at 708fbc70ce4695690f2cd04aad9c92317c705a7c

Files

T

rpotter6298 708fbc70ce Add analysis scripts and experiment configurations for bridge attention and sensitivity studies

- Introduced `bridge_attention_ceiling_check.py` for variance decomposition analysis on bridge attention configurations.
- Added `bridge_attention_readout.py` to perform per-tower gate and contribution readouts, including AUC sanity checks.
- Created multiple JSON configuration files for backbone replication experiments, including anonymous CV variants and basic backbones.
- Implemented sensitivity experiments to evaluate the impact of axial length inclusion and EfficientNetV2-M performance at higher resolutions.
- Added a memory probe script to assess GPU memory usage during training with EfficientNetV2-M.

2026-07-03 08:51:44 +02:00

647 B

Raw Blame History

1	feature	mean_drop	std_drop	baseline_auc_mean
2	Age	0.1487745098039216	0.07246596166317897	0.7117647058823529
3	IOP_corr	0.0645588235294118	0.053595964337907635	0.7117647058823529
4	Phakic/Pseudophakic	0.031004901960784353	0.05039528009700277	0.7117647058823529
5	Pachymetry	0.011421568627451003	0.016547911692771797	0.7117647058823529
6	Gender	0.006102941176470622	0.022596178307964714	0.7117647058823529
7	eyeID	0.0	0.0	0.7117647058823529
8	dioptre_2	-0.0010294117647058861	0.003622645200240852	0.7117647058823529
9	astigmatism	-0.002156862745098008	0.008336072214271729	0.7117647058823529
10	dioptre_1	-0.006593137254901939	0.011003619078899692	0.7117647058823529

647 B Raw Blame History

647 B

Raw Blame History