Abstract:This paper presents an interactive platform to interpret multi-objective evolutionary algorithms. Sokoban level generation is selected as a showcase for its widespread use in procedural content generation. By balancing the emptiness and spatial diversity of Sokoban levels, we illustrate the improved two-archive algorithm, Two_Arch2, a well-known multi-objective evolutionary algorithm. Our web-based platform integrates Two_Arch2 into an interface that visually and interactively demonstrates the evolutionary process in real-time. Designed to bridge theoretical optimisation strategies with practical game generation applications, the interface is also accessible to both researchers and beginners to multi-objective evolutionary algorithms or procedural content generation on a website. Through dynamic visualisations and interactive gameplay demonstrations, this web-based platform also has potential as an educational tool.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: how to make the optimization process of multi - objective evolutionary algorithms (MOEAs) visual and easy to understand through an interactive platform. Specifically, the author chooses the Sokoban level generation as an example to show how to balance the emptiness and spatial diversity of the levels, thereby explaining the working principle of multi - objective evolutionary algorithms and their performance in practical applications.
### Problem Background
Multi - objective optimization problems (MOPs) are widely present in various fields, such as game optimization, aero - engine calibration, trauma system design, and bias mitigation in decision - making. Multi - objective evolutionary algorithms (MOEAs) perform well in dealing with these complex problems and can generate diverse and high - quality solutions. However, for those who are not familiar with MOEAs, understanding and studying these algorithms can be challenging.
### Main Contributions of the Paper
1. **Developed a web - based interactive platform**: This platform can visually display the optimization process of MOEAs in real - time, helping users better understand the working mechanism of the algorithm.
2. **Selected Sokoban as a case**: By generating Sokoban levels, it shows how to balance the two conflicting goals of emptiness and spatial diversity in level design.
3. **Integrated the Two Arch2 algorithm**: This is a well - known multi - objective evolutionary algorithm that can maintain two archives (the convergence archive CA and the diversity archive DA) during the optimization process, thus ensuring the diversity and quality of solutions.
4. **Provided the function of an educational tool**: Through dynamic visualization and interactive demonstrations, this platform is suitable not only for researchers but also for beginners to learn MOEAs and procedural content generation (PCG).
### Mathematical Formulas
In Sokoban level generation, the optimization goals are defined as maximizing emptiness and spatial diversity:
\[
\max \{ f_{\text{emp}}(\text{level}), f_{\text{div}}(\text{level}) \}
\]
where:
- \( f_{\text{emp}} \) represents the proportion of empty spaces in the level, that is:
\[
f_{\text{emp}}=\frac{\text{number of empty spaces}}{\text{total number of grids}}
\]
- \( f_{\text{div}} \) represents spatial diversity, and the calculation formula is as follows:
\[
f_{\text{div}} = -\frac{1}{\log n} \sum_{i = 1}^{n} \alpha p_i \log \left( \frac{\alpha p_i}{n} \right)
\]
where:
- \( n \) is the number of parts into which the level is divided;
- \( p_i \) is the proportion of empty spaces in the \( i \) - th part;
- \( \alpha \) is a normalization factor, ensuring that \( \sum_{i = 1}^{n} \alpha p_i=1 \).
Through these formulas, the platform can evaluate each generated level and continuously improve during the optimization process.
### Summary
This paper solves the problem of MOEAs being difficult to understand and apply by constructing an interactive platform. In particular, through the specific example of Sokoban level generation, it shows how to balance multiple optimization goals in practical applications and provides an intuitive teaching tool.