Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Review on Single-Problem Multi-Attempt Heuristic Optimization

Judith Echevarrieta, Etor Arza, Aritz Pérez, Josu Ceberio

TL;DR

This paper addresses Single-Problem Multi-Attempt Heuristic Optimization (SIMHO), where solving a single optimization problem is pursued through multiple deliberate attempts using different heuristics or configurations. It introduces a unified SIMHO framework and an abstract strategy that alternates information generation and distribution updates, formalizing the problem as selecting the best performing alternative from a set $A$ based on observed outcomes $I_a$ and quality $f_a$. The authors survey four topics—Algorithm Selection, Parameter Tuning, Multi-Start, and Resource Allocation—and present a taxonomy of updating criteria and adaptive probability meanings to classify existing work within the framework. They argue the framework enables design of new strategies and a software library for SIMHO, and identify opportunities to extend to strategies that jointly select more components for greater practical impact. Overall, the framework provides a cohesive, extensible lens to integrate and develop SIMHO strategies across the four components of an alternative, guiding practical deployment and future research.

Abstract

In certain real-world optimization scenarios, practitioners are not interested in solving multiple problems but rather in finding the best solution to a single, specific problem. When the computational budget is large relative to the cost of evaluating a candidate solution, multiple heuristic alternatives can be tried to solve the same given problem, each possibly with a different algorithm, parameter configuration, initialization, or stopping criterion. The sequential selection of which alternative to try next is crucial for efficiently identifying the one that provides the best possible solution across multiple attempts. Despite the relevance of this problem in practice, it has not yet been the exclusive focus of any existing review. Several sequential alternative selection strategies have been proposed in different research topics, but they have not been comprehensively and systematically unified under a common perspective. This work presents a focused review of single-problem multi-attempt heuristic optimization. It brings together suitable strategies to this problem that have been studied separately through algorithm selection, parameter tuning, multi-start and resource allocation. These strategies are explained using a unified terminology within a common framework, which supports the development of a taxonomy for systematically organizing and classifying them.

A Review on Single-Problem Multi-Attempt Heuristic Optimization

TL;DR

This paper addresses Single-Problem Multi-Attempt Heuristic Optimization (SIMHO), where solving a single optimization problem is pursued through multiple deliberate attempts using different heuristics or configurations. It introduces a unified SIMHO framework and an abstract strategy that alternates information generation and distribution updates, formalizing the problem as selecting the best performing alternative from a set based on observed outcomes and quality . The authors survey four topics—Algorithm Selection, Parameter Tuning, Multi-Start, and Resource Allocation—and present a taxonomy of updating criteria and adaptive probability meanings to classify existing work within the framework. They argue the framework enables design of new strategies and a software library for SIMHO, and identify opportunities to extend to strategies that jointly select more components for greater practical impact. Overall, the framework provides a cohesive, extensible lens to integrate and develop SIMHO strategies across the four components of an alternative, guiding practical deployment and future research.

Abstract

In certain real-world optimization scenarios, practitioners are not interested in solving multiple problems but rather in finding the best solution to a single, specific problem. When the computational budget is large relative to the cost of evaluating a candidate solution, multiple heuristic alternatives can be tried to solve the same given problem, each possibly with a different algorithm, parameter configuration, initialization, or stopping criterion. The sequential selection of which alternative to try next is crucial for efficiently identifying the one that provides the best possible solution across multiple attempts. Despite the relevance of this problem in practice, it has not yet been the exclusive focus of any existing review. Several sequential alternative selection strategies have been proposed in different research topics, but they have not been comprehensively and systematically unified under a common perspective. This work presents a focused review of single-problem multi-attempt heuristic optimization. It brings together suitable strategies to this problem that have been studied separately through algorithm selection, parameter tuning, multi-start and resource allocation. These strategies are explained using a unified terminology within a common framework, which supports the development of a taxonomy for systematically organizing and classifying them.

Paper Structure

This paper contains 19 sections, 5 equations, 1 figure, 6 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related work
  3. SIMHO framework
  4. Multi-attempt heuristic optimization
  5. The abstract SIMHO strategy
  6. Topics addressing SIMHO
  7. Algorithm Selection
  8. Parameter Tuning
  9. Multi-Start
  10. Resource Allocation
  11. Taxonomy of SIMHO strategies
  12. Updating criteria
  13. Quality meaning of probability in adaptive criterion
  14. Two-stage process for adaptive criterion
  15. Probability related to expected maximum quality
  16. ...and 4 more sections

Figures (1)

  • Figure 1: The first box (named Single-problem Multi-attempt Heuristic Optimization framework...) highlights in yellow the contributions of a review carried out throughout the proposed framework. The second box under it (named ...together with a taxonomy of selection strategies) enumerates in purple the linked classification rules in blue of a SIMHO strategy, with the categories defined for each of them in green, and the contributions implied by the resulting taxonomy in yellow.