update README

.gitignore

@@ -0,0 +1,150 @@
+*.npy
+*.csv
+*.pdf
+old/
+experiments/
+results/
+res/
+logs/
+data/
+pkls/
+test/
+scrub/trained_models
+*.ipynb
+
+.DS_Store
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+#george's ignores
+**/*.swp
+**/__pycache__
+kernels/

README.md

@@ -1,2 +1,4 @@
-# LCODEC_Deep_Unlearning_CVPR22
-Code for CVPR22 paper "Deep Unlearning via Randomized Conditionally Independent Hessians"
+#  Deep Unlearning via Randomized Conditionally Independent Hessians (CVPR 2022)
+
+All experiments are run within the specified folders, and call out to 'codec'.
+Navigate to each folder for example scripts and directions on how to run in __expname__/README.md.

bullseye/README.md

@@ -0,0 +1,4 @@
+To run Bullseye comparisons, navigate to model-augmented-mutual-information and run `roc_gen.sh'.
+For feature mappings, uncomment line 75 in roc_gen.py to enable training mapping.
+
+Note: We have adapted code for the paper: [Model-Augmented Conditional Mutual Information Estimation for Feature Selection (UAI 2020)](https://github.com/syanga/model-augmented-mutual-information) and also [pycit](https://github.com/syanga/pycit). Hence, we have added the modified sources here.

bullseye/model-augmented-mutual-information-master/.gitignore

@@ -0,0 +1,134 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+.vscode/
+data/
+.DS_Store
+saved/
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/

bullseye/model-augmented-mutual-information-master/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2020 Alan Yang
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

bullseye/model-augmented-mutual-information-master/README.md

@@ -0,0 +1,39 @@
+# model-augmented-mutual-information
+
+This repository contains code related to the paper "Model-Augmented Conditional Mutual Information Estimation for Feature Selection," which appeared at the 2020 Conference on Uncertainty in Artificial Intelligence (UAI) [[Link]](https://arxiv.org/abs/1911.04628).
+
+The code in this repository depends on [```pycit```](https://github.com/syanga/pycit), which can be installed using pip:
+```
+pip install pycit
+```
+
+
+# Experiments
+
+## 1. Improving the Performance of k-NN Mutual Information Estimator 
+
+This experiment shows how a learned mapping can improve the performance of the k-NN mutual information (MI) estimator. We have a random variable X which is sampled uniformly on two concentric rings; a scatterplot of 500 samples is shown below:
+
+<img src="assets/bullseye.png" width="200">
+
+Let Y be the magnitude of X plus noise. The goal is to estimate I(X;Y), the MI between X and Y. The samples of X are spread across two rings instead of just along a single axis, which makes k-NN MI estimation more difficult. Therefore, before using the k-NN estimator, we would like to learn a mapping of X that looks more like:
+
+<img src="assets/mapping_regularized.png" width="200">
+
+as opposed to:
+
+<img src="assets/mapping_nominal.png" width="200">
+
+We learn this mapping with a regularization term that maps values of X with similar information about Y (in this case, similar magnitudes) close together. This is done in the IPython notebook [```bullseye2d_experiment.ipynb```](bullseye2d_experiment.ipynb).
+
+
+## 2. Conditional Independence Testing
+
+In this experiment, we evaluate the performance of the k-NN based conditional independence test (CIT) on a 3D version of the Bullseye data, where the variables' distribution is faithful to the DAG structure:
+
+<img src="assets/dag7.png" width="200">
+
+Here, the features are 3-dimensional, and the target variable Y is a scalar. The Markov blanket of Y is highlighted in blue. We find that the k-NN test is not able to correctly find the Markov blanket (parents, children, and co-parents of Y) using a PC-type algorithm (see [```pycit```](https://github.com/syanga/pycit) documentation, and Algorithm 2 in our [paper](https://arxiv.org/abs/1911.04628)). However, using samples of a learned feature mapping allowed the k-NN method to succeed. The code for this experiment is in the IPython notebook [```bullseye3d_experiment.ipynb```](bullseye3d_experiment.ipynb).
+
+
+

bullseye/model-augmented-mutual-information-master/bullseye/__init__.py

@@ -0,0 +1,3 @@
+""" Generate bullseye data """
+from .bullseye import *
+from .bullseye_network import *