This file is intended to document the performance of the various models. Typically, models are measured using 10-fold cross validation, and storing the predictions made in each fold.
To regenerate the images in this report, run:
make
make -f Makefile.model clean
make -j1 -f Makefile.model
This is an XGBoost model based on age and gender metadata.
This is a single-layer network using only the first frame of each 2 chamber or 4 chamber view. This is equivalent to linear regression using the scaled pixel data as explanatory variables.
This model is trained on all 30 frames of each 4 chamber view.
