Landscape Analysis of Optimization Problems and Algorithms
Description
The notion of a fitness landscape was first introduced in 1932 to understand natural evolution. Still, the concept was later applied in the context of evolutionary computation to understand algorithm behaviour on different problems. Over the last decade, the field of fitness landscapes has experienced a significant surge in research, as evidenced by the increasing number of published papers on the topic, as well as regular tutorials, workshops, and special sessions at major evolutionary computation conferences. More recently, landscape analysis has been applied in contexts beyond evolutionary computation in areas such as neural architecture search, feature selection for data mining, hyperparameter optimisation and neural network training.
A further recent advance has been the analysis of landscapes through the trajectories of algorithms. The search paths of algorithms provide samples of the search space that can be interpreted as a view of the landscape from the algorithm's perspective. What algorithms "see" as they move through the search space of different problems can help us understand how evolutionary and other search algorithms behave on problems with different characteristics.
The tutorial covers foundational work in fitness landscape analysis as well as new techniques developed in the last few years. We will cover recent results on Pareto local optima solution networks (PLOS-nets), an extension of local optima networks to model multi-objective fitness landscapes. We also cover search trajectory networks (STNs) as a tool to analyse and visualise the behaviour of metaheuristics through modelling the search trajectories of algorithms. With PLOS-nets and STNs, we can visualise realistic search spaces in ways not previously possible and bring a whole new set of quantitative network metrics for characterising and understanding computational search.
Case studies will be presented of recent applications of landscape analysis in both discrete and continuous domains. These include insights into real-world optimisation problems, landscape-aware constraint handling, training of neural networks, and Pseudo-boolean multi-objective optimisation. For search trajectory networks, applications in combinatorial optimisation, neuroevolution and genetic programming will be discussed.
Organizers
Gabriela Ochoa is a Professor of Computing Science at the University of Stirling in Scotland, UK. Her research lies in the foundations and methods of evolutionary algorithms and metaheuristics, with an emphasis on fitness landscape analysis and visualisation, grey-box optimisation, autonomous search, and cross-disciplinary applications in healthcare. She holds a PhD from the University of Sussex, UK, and has worked at the University Simon Bolivar, Venezuela, and the University of Nottingham, UK, before joining Stirling. Her Google Scholar h-index is 47. She has been instrumental in creating and popularising the concepts of local optima networks (LONs) and search trajectory networks (STNs). She was the EiC for the Genetic and Evolutionary Computation Conference (GECCO) in 2017 and 2025, and belongs to the editorial boards of the Evolutionary Computation Journal (ECJ) and the ACM Transactions on Evolutionary Learning and Optimisation (TELO). She is a member of the executive board of the ACM interest group in evolutionary computation (SIGEVO), where she edits the SIGEVOlution newsletter, and of the SPECIES society (organising EvoStar conferences). In 2020, she was recognised by the leading European event on bio-inspired algorithms, EvoStar, for her outstanding contributions to the field.