This is the accompanying code for the paper: NEU: A Meta-Algorithm for Universal UAP-Invariant Feature Representation

by: A. Kratsios and C. Hyndman

Published: Journal of Machine Learning Research (2021) Volume: 22 Pages: 1-52

Available online at: https://jmlr.org/papers/v22/18-803.html

@article{JMLR:v22:18-803,
         author  = {Anastasis Kratsios and Cody Hyndman},
         title   = {NEU: A Meta-Algorithm for Universal UAP-Invariant Feature Representation},
         journal = {Journal of Machine Learning Research},
         year    = {2021},
         volume  = {22},
         number  = {92},
         pages   = {1-51},
         url     = {http://jmlr.org/papers/v22/18-803.html}
        }

NEU is a meta-procedure which generically learns a robust feature map which upgrades the performance of standard learning algorithms.

• An implementation of NEU-OLS, NEU-PCA, and NEU Gradient boosted random forest,

• An implementation of NEU-PCA and NEU-Auto-Encoder

• A data simulator to test NEU and its benchmarked methods against.