Non-Parametric Dimensionality Reduction using Entropy via NEST algorithm
Primary Investigator:
Hany Abdel-Khalik
Tyler Lewis
Abstract
This project is mainly concerned with surrogate data generation for use in application such as anomaly detection, and design optimization, which allow for lower uncertainty and robustness to noise. Surrogate data methodologies currently in use often rely on analyst intuition which risks design bias in the project. NEST proposes an entropy-based detrending algorithm to avoid such bias and produce random errors which increase the efficacy of surrogate data.