From Performance to Understanding: A Vision for Explainable Automated Algorithm Design

Niki van Stein, Anna V. Kononova, Thomas Bäck · 2025

Read From Performance to Understanding: A Vision for Explainable Automated Algorithm Design by Niki van Stein, Anna V. Kononova, Thomas Bäck (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-16201.txt.

Description

Niki van Stein, Anna V. Kononova, and Thomas Bäck argue that automated algorithm design needs explanation, not just performance. Large language models can now generate full optimization heuristics. They can also search vast design spaces and adapt through feedback. Yet that progress is opaque. Rarely does anyone learn why a generated algorithm works, which parts matter, or how choices relate to problem structure.

Their vision rests on three pillars. First, language-model-driven discovery makes algorithmic variants. Second, explainable benchmarking attributes performance to parts and settings. Third, problem-class descriptors connect algorithm behavior to landscape structure. Together they form a closed knowledge loop where discovery, explanation, and generalization reinforce each other, shifting the field toward interpretable, class-specific design.

Catalog record

CREATORNiki van Stein, Anna V. Kononova, Thomas Bäck
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-16201

Browse the full library · About the Tower of Babel