Run generalte_2D.py and then train.py to play with the toy model of BIN. Alternatively you could run the script run.sh directly in one click.
This repo supports the illustration of our idea on 2D classification and regression, 3D regression.
The results of 3D classification toymodel could not fully illustrate our idea, because the landscape of the 3D classification model itself is smoothing enough after training.