This is the code used to generate the results for the Neurips 2020 conference paper "Optimal Learning from Verified Training Data".

This code has the following requirements:
- MATLAB 2019B or later
- CVX tool (http://cvxr.com/cvx/)
- The MATLAB Statistics and Machine Learning Toolbox (https://uk.mathworks.com/products/statistics.html)
- The MATLAB Parallel Computing Toolbox (https://uk.mathworks.com/products/parallel-computing.html)
- The MATLAB Optimization Toolbox (https://uk.mathworks.com/products/optimization.html)

To run the experiments used to plot the figures in both the main paper and the supplementary material, run sdp_results.m.
To plot results corresponding to the red wine dataset use plot_agg_wine.m, whilst to plot results for the medical personal
costs dataset run plot_agg_in. To plot runtime results run plot_time_agg.m.

If you encounter any issues with running this code, please contact Nicholas Bishop (nb8g13@soton.ac.uk).