facebookresearch/fbcdgraph

The codes reproduce the figures and statistics in the paper, "Cumulative deviation of a subpopulation from the full population," by Mark Tygert. The repo also provides the LaTeX and BibTex sources required for replicating the paper.

The accompanying codes reproduce all figures and statistics presented in "Cumulative deviation of a subpopulation from the full population" by Mark Tygert. This repository also provides the LaTeX and BibTeX sources required for replicating the paper.

The main files in the repository are the following:

./paper.pdf PDF version of the paper

./paper.tex LaTeX source for the paper

./paper.bib BibTeX source for the paper

codes/calibr.py Functions for plotting calibration, both cumulative and reliability diagrams

codes/calibr_weighted.py Functions for plotting calibration with weighted samples

codes/subpop.py Functions for plotting deviation of a subpop. from the full pop.

codes/subpop_weighted.py Functions for plotting deviation of a subpop. from the full pop. with weighting

codes/imagenet.py Python script for processing ImageNet using a pre-trained ResNet-18

codes/imagenet_classes.txt Text file containing a dictionary of the names of the classes in ImageNet

codes/acs.py Python script for processing the American Community Survey

codes/psam_h06.csv Microdata from the 2019 American Community Survey of the U.S. Census Bureau

Regenerating all the figures requires running in the directory codes every Python file there. All but codes/imagenet.py need only CPUs, whereas codes/imagenet.py runs on a GPU simultaneously with running on CPU cores.

---

License

This fbcdgraph software is licensed under the LICENSE file (the MIT license) in the root directory of this source tree.