SOS

SOS implements statistical benchmarking, parameter tuning, and testbed management for stochastic optimisation research by providing implementations of statistical tests, an IEEE CEC 2014-derived continuous testbed with controllable rotation operators, and tools for data collection and algorithmic overhead measurement.

Key Features:

• Parameter tuning and algorithm comparison: Supports automated parameter tuning workflows and the implementation of comparison algorithms for metaheuristics.
• Statistical tests: Provides tailored implementations of the Wilcoxon Rank-Sum test and the Holm–Bonferroni procedure with result export to PDF and LaTeX.
• Advanced pairwise comparison routine: Includes an original advanced statistical routine specifically for accurate pairwise comparison of stochastic optimisation algorithms.
• Continuous testbed suite: Offers a novel testbed derived from the IEEE CEC 2014 benchmark with controllable activation of rotation operators across test functions.
• Data collection and overhead measurement: Collects and processes experimental data and measures algorithmic overhead for performance analysis.
• Implementation: Implemented as a Java-based platform for research-oriented execution of the above features.

Scientific Applications:

• Benchmarking metaheuristics: Statistical benchmarking and rigorous performance comparison of stochastic optimisation algorithms.
• Parameter optimization: Systematic parameter tuning of metaheuristic algorithms to assess sensitivity and performance.
• Controlled continuous optimization experiments: Evaluation of algorithms on IEEE CEC 2014-derived continuous functions with configurable rotation operators.
• Pairwise algorithm comparison: Accurate pairwise statistical comparisons of stochastic optimisation algorithms using specialized routines.
• Performance profiling: Measurement of algorithmic overhead and generation of processed experimental results for reporting.

Methodology:

Implements the Wilcoxon Rank-Sum test and Holm–Bonferroni procedure, an original advanced statistical routine for pairwise comparisons, an IEEE CEC 2014-derived continuous testbed with controllable rotation operators, experimental data collection and processing, algorithmic overhead measurement, parameter tuning and algorithm comparisons, and result export to PDF and LaTeX.