Skip to content
Code to replicate the experimental results in paper "Adversarial Regression with Multiple Learners" @ ICML 2018
Python Shell
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Adversarial Regression with Multiple Learners


Currently, you can use this code to replicate the experimental results for redwine dataset (Figure 1, Figure 2, and Figure 7). The experimental results for other datasets can be similarly generated.


(Necessary: Python3.7 and conda)

  1. First, clone the project folder to your computer.
  2. Then, create an environment and activate it:
conda create -n multiple-learner python=3.7
conda activate multiple-learner
  1. After the environment is activated, install the following required packages:
    conda install numpy scipy pandas scikit-learn seaborn matplotlib
    pip install cvxpy
    pip install cvxopt


  1. Inside the project folder, create a folder to store experimental outputs:
mkdir result/
  1. Enter into src/ folder, run the following command to generate experimental outputs:
  1. Insider src/ folder, run the following command to generate Figure 1 (complete information), Figure 2 (incomplete information + over-estimatd z), and Figure 3 (incomplete information + under-estimated z):
python redwine
  1. I generated the figures on MacOS Mojave. If you see some error like the following, Try this solution:
libc++abi.dylib: terminating with uncaught exception of type NSException


  title={Adversarial Regression with Multiple Learners},
  author={Tong, Liang and Yu, Sixie and Alfeld, Scott and others},
  booktitle={International Conference on Machine Learning},
You can’t perform that action at this time.