Deep-Bias

By Bas van Stein et al
Read the original document by opening this link in a new tab.

Table of Contents

1 Introduction
2 Structural bias
2.1 Existing methodology for measuring SB
2.2 BIAS toolbox
3 Methodology
3.1 Portfolio of scenarios
3.2 Optimized Convolutional Neural Network
3.3 Predicting SB and SB type
4 Benchmarking structural bias of real algorithmic data
4.1 Data collection setup

Summary

The document discusses the Deep-Bias method for detecting structural bias using explainable AI. It introduces a deep-learning expansion of the BIAS toolbox called Deep-Bias, which outperforms the original toolbox in detecting and classifying bias in heuristic optimization algorithms. The methodology includes a portfolio of scenarios, an optimized convolutional neural network, and prediction of structural bias and its types. The Deep-Bias method is compared to the original BIAS toolbox, showing slightly better performance. It further benchmarks structural bias in real algorithmic data of 336 optimization heuristics.
×
This is where the content will go.