Notices for the Clover project

This product includes software from the Private PGM project, under the license Apache version 2.0.
https://github.com/ryan112358/private-pgm
The file mst.py in the folder mechanisms was modified in the following ways:
    - The parameter num_samples was added in the MST function
    - Only the functions MST, measure, compress_domain, exponential_mechanism, select, transform_data and reverse_data
      are included. Other elements of the code were removed.
    - The paths of imported modules were modified to be relative to the location of this file.

This product includes software from the Synthpop project, under the MIT license.
https://github.com/hazy/synthpop
The following changes were made to the code:
    - The paths of imported modules were modified to be relative.
    - Replace iteritems() by items() in validator.py and synthpop.py (deprecated)
    - Add **kwargs / self.args parameter in synthpop.py to pass arguments to the generation method
    - Add max_depth parameter in the decision tree classifier/regressor in cart.py

This product includes software from the SDV (and CTGAN) project, under the Business Source License 1.1.
The Licensed Work is (c) DataCebo, Inc.
https://github.com/sdv-dev
The following changes were made to the code:
    - The paths of imported modules were modified.
    - The fitting methods were modified to accommodate for differentially private training.
    - In CTGAN, the gradient penalty was removed in the DP version as the gradient clipping respects the imposed
      constraints (clipping norm of 1).

This product includes software from the TabDDPM project, under the MIT license.
https://github.com/yandex-research/tab-ddpm

This product includes software from the CTAB-GAN-Plus project, under the license Apache version 2.0.
https://github.com/Team-TUD/CTAB-GAN-Plus

This product includes software from the dpart project, under the MIT license.
https://github.com/hazy/dpart
The following changes were made to the code:
    - The paths of imported modules were modified to be relative.
    - Modify the differential private synthpop class to include more arguments.
    - Implement privacy budget accountant.
    - Replace the lambda function in the dpart class by a named function so that the object can be saved with pickle.
    - Change feature_range of the MinMaxScaler from list to tuple to match the type required by sklearn.
    - Convert the type of the columns to str type when generating synthetic data.
    - Update maximum number of iterations for logistic regression to fix the convergence issue.
    - Set the entropy to its one-sided limitation when the probability is 0 to avoid returning NA.

Please refer to the LICENSE file for details about the licensing of this project.