Jul 2026· Proceedings of the Genetic and Evolutionary Computation Conference Companion· 1 citation· 14 references
Computer Science
Abstract
Large Language Models (LLMs) are opening new directions for automated heuristic design (AHD), allowing evolutionary methods to create and enhance heuristics for constrained optimization problems (COPs). However, most existing approaches face the challenge of the exploration-exploitation balance, where the evolution needs to escape convergence to homogeneous populations and discover as large a heuristic landscape as possible. To address this challenge, we introduce Quality-Diversity Evolution (QDEvo), a multi-objective framework that integrates Quality-Diversity optimization with LLM-based AHD. At its core, QDEvo employs a semantic survival selection mechanism that clusters algorithms by functional similarity, then applies local Pareto competition. Evaluation on well-known COPs benchmarks and real-world problems shows that our method consistently outperforms the state-of-the-art baseline in both Hypervolume and Inverted Generational Distance metrics. These results facilitate further exploration of the algorithmic design space, while ensuring competitive solution quality and efficiency.
This work proposes a general LES framework that incorporates feature-assisted niche construction within abstract search spaces, enabling the seamless integration of niche-based search strategies from evolutionary computation and introduces PartEvo (Partition to Evolve), an LES method that combines niche collaborative search and advanced prompting strategies to improve algorithm discovery efficiency.
Qinglong Hu, Qingfu Zhang· Neural Information Processin...· 9 citations· ⚡3
The number and arrangement of local optima are crucial for evaluating the exploration-exploitation balance and the ability to discover promising local optima. However, existing benchmark suites rarely provide explicitly defined multi-modal landscapes, especially in multi-objective optimization. To bridge this gap, we propose a framework for designing both single- and multi-objective benchmark problems with identifiable local optima and controllable landscape features. Our approach builds on Max-Set of Gaussians (MSG) landscapes and extends them to the multi-objective domain. To match target landscape features, we optimize Gaussian heights and variances. Empirical results demonstrate that the framework can emulate several function classes from the Black-Box Optimization Benchmarking (BBOB) suite, including separable, multimodal, and ill-conditioned functions. Furthermore, we confirm that the framework can generate problems with Exploratory Landscape Analysis (ELA) feature patterns that are not present in the BBOB suite. Our results also imply that the landscape features of single-objective MSG landscapes are inherited in multi-objective extensions.
Shu Tanaka, Shoichiro Tanaka, Kippei Mizuta et al.· Annual Conference on Genetic...· 0 citations
In many real-world domains, optimization problems involve both quantitative objectives and qualitative criteria that cannot be directly formulated as mathematical functions. While recent research has integrated large language models (LLMs) into optimization algorithms, these approaches primarily use LLMs to guide the search process, generate heuristics, or tune optimization strategies, rather than to evaluate semantic objectives. This paper introduces generative heuristics, a methodology that combines traditional metaheuristics with LLM-based semantic evaluation to address ‘soft optimization’ problems containing both hard quantitative constraints and soft qualitative objectives. The proposed methodology first applies a metaheuristic algorithm to generate a shortlist of high-quality candidate solutions satisfying the quantitative objectives. These candidates are then evaluated by an LLM according to a user-defined qualitative rubric, and the resulting semantic scores are incorporated into the optimization process through a weighted soft objective. The methodology is illustrated through a constrained portfolio optimization problem in which financial risk is minimized while simultaneously promoting strategic objectives related to sector diversification, geographic exposure, environmental-social-governance quality, and overall portfolio coherence. Experimental results show that the proposed methodology consistently improves strategic alignment while producing solutions that remain close to those obtained from the hard optimization formulation.
Miguel Saiz, A. Juan, J. Panadero· Mathematics· 0 citations
Rule-based classifiers offer high interpretability, which is essential in domains such as medicine where decision transparency is key. However, multi-objective rule learning remains a complex challenge due to the highly combinatorial nature of the search space. This work investigates the hybridization of a Genetic Algorithm (GA) with Pareto Local Search (PLS) to improve the exploration of non-dominated rulesets. Built upon the MH-Builder meta-heuristics framework, and the existing Multi-Objective Classification Algorithm for Imbalanced Data, or MOCA-I, our proposed pipeline uses a GA phase for global exploration followed by an Iterated PLS phase for intensification. Our results demonstrate that while the hybrid MOCA-I+GA pipeline maintains the same high predictive performance as MOCA-I, the evolutionary phase provides only marginal improvements in diversity (Hypervolume and Spread) at a significant computational cost. This study provides a documented baseline for hybrid metaheuristics in rule mining and discusses the limitations of loose-coupled sequential hybridization.
Evgueni Blanquart, L. Jourdan, Nadarajen Veerapen· Proceedings of the Genetic a...· 0 citations
Designing effective multi-objective Bayesian optimization (MOBO) algorithms requires balancing many interdependent design choices whose optimal configuration is problem-dependent and typically demands deep expertise. We extend the LLaMEA framework to MOBO, using large language models as mutation and crossover operators within evolutionary strategies to generate complete algorithm implementations, with SMAC hyperparameter optimization integrated into the evolutionary loop. Across nine evolutionary runs we generated approximately 900 algorithms and benchmarked them on twelve synthetic problems (ZDT, DTLZ, WFG) and three real-world engineering problems (RE), using a BoFire qParEGO implementation as a state-of-the-art Bayesian-optimization baseline. On the synthetic suite the strongest generated algorithm attains the highest mean normalized hypervolume (0.971, vs. 0.869 for qParEGO) while requiring roughly 60x less wall-clock time; a Friedman test with post-hoc analysis places the two in a single top-performing group, and per-problem tests find the generated algorithm significantly better than qParEGO on 7 of the 12 problems and never worse, matching state-of-the-art accuracy at an order-of-magnitude lower cost. On the three unseen real-world engineering problems a generated algorithm attains the best mean normalized hypervolume (0.985, vs. 0.971 for qParEGO)--significantly better than qParEGO on two of the three problems--at roughly 3.4x lower wall-clock cost, confirming that the gains transfer beyond the synthetic regime. LLM-driven evolutionary search can thus discover algorithm designs that achieve Pareto-efficient trade-offs difficult to reach through manual design.
G. Laskaris, R. Brasher, N. V. Stein et al.· 0 citations
Nested Tournament (NT) is a Multi-Objective (MO) selection method that enables fine-grained control of selection pressure through sequential single-objective tournaments. Although previously proposed, the impact of elitism strategies within NT remains largely unexplored. This study systematically investigates multiple elitism mechanisms for NT within tree-based MO Genetic Programming (MOGP), including NSGA-II population replacement, crowding distance, first-objective, and a novel ideal-point strategy, comparing them against non-elitist NT and standard NSGA-II. Experiments are conducted with up to five objectives for the accuracy-complexity trade-off and show that elitism design critically influences stability, efficiency, and semantic diversity. Notably, simpler NT-specific elitism strategies achieve comparable performance to NSGA-II at lower computational cost while better preserving semantic diversity. Overall, the findings highlight NT as an efficient MOGP alternative to selection methods based on Pareto Fronts.
Filipa Vieira Goncalves Pereira, Karina Brotto Rebuli, M. Giacobini et al.· Proceedings of the Genetic a...· 0 citations