Invited talk of Thomas Bäck at the 20th International Conference on Unconventional Computing and Natural Computation

On March 13, Thomas Bäck gave an invited talk at the UCNC 2023 conference in Jacksonville, Florida. This conference brings together scientists interested in novel forms of computation, computation inspired by nature, and computational aspects of natural processes. The presentation is entitled “On the Automatic Optimization of Problem-Specific Optimization Heuristics Gleaned from Nature”.

The abstract of the presentation is as follows:

For decades, researchers have been looking at paradigms gleaned from nature as inspiration for problem solving approaches, for example in the domain of optimization. There are many classes of such algorithms, including for example evolutionary algorithms, particle swarms, differential evolution, ant colony optimization, and the number of proposed variants of them is quite large. This makes it hard to keep track of the variants and their respective strengths, and even more so it creates a difficult situation for non-experts who are interested in selecting the best algorithm for their real-world application problem.

In this presentation, I propose the idea to automatically optimize the optimization heuristic. This task can be approached as an algorithm configuration problem, for which I will present some examples illustrating that this task can be handled by direct global optimization algorithms – in other words, by “automatically optimizing the optimization algorithm”. I will give an example how a combinatorial design space of 4608 configuration variants of evolution strategies can be searched, and how the results can be analyzed using data mining. This approach provides an opportunity for discovering the unexplored areas of the optimization algorithm design space. Extensions towards other algorithm design spaces such as particle swarm optimization and differential evolution are then outlined, too.

In the second part of the presentation, I will discuss a range of real-world engineering design applications, for which such an approach could truly provide a competitive advantage. In such cases, optimizing the optimization algorithm requires a proper definition of the problem class, for which the optimization is executed. For the example of automotive crash optimization problems, I will present first results demonstrating that these problems differ a lot from the classical benchmark test function sets used by academic community, and present an automated approach to find test functions that properly represent the real-world problem. First results on the performance gain that can be achieved by optimizing the optimization algorithm on such real-world problems are also presented.