Search

Population based incremental learning algorithms and selection hyper-heuristics are highly adaptive methods which can handle different types of dynamism that may occur while a given problem is being solved. In this study, we present an approach based on a multi-population framework hybridizing these methods to solve dynamic environment problems. A key feature of the hybrid approach is the utilization of offline and online learning methods at successive stages. The performance of our approach along with the influence of different heuristic selection methods used within the selection hyper-heuristic is investigated over a range of dynamic environments produced by a well known benchmark generator as well as a real world problem, referred to as the Unit Commitment Problem. The empirical results show that the proposed approach using a particular hyper-heuristic outperforms some of the best known approaches in literature on the dynamic environment problems dealt with. [ABSTRACT FROM AUTHOR]

Deep neural networks have significantly advanced medical image classification across various modalities and tasks. However, manually designing these networks is often time-consuming and suboptimal. Neural Architecture Search (NAS) automates this process, potentially finding more efficient and effective models. This study provides a comprehensive comparative analysis of our two NAS methods, PBC-NAS and BioNAS, across multiple biomedical image classification tasks using the MedMNIST dataset. Our experiments evaluate these methods based on classification performance (Accuracy (ACC) and Area Under the Curve (AUC)) and computational complexity (Floating Point Operation Counts). Results demonstrate that BioNAS models slightly outperform PBC-NAS models in accuracy, with BioNAS-2 achieving the highest average accuracy of 0.848. However, PBC-NAS models exhibit superior computational efficiency, with PBC-NAS-2 achieving the lowest average FLOPs of 0.82 GB. Both methods outperform state-of-the-art architectures like ResNet-18 and ResNet-50 and AutoML frameworks such as auto-sklearn, AutoKeras, and Google AutoML. Additionally, PBC-NAS and BioNAS outperform other NAS studies in average ACC results (except MSTF-NAS), and show highly competitive results in average AUC. We conduct extensive ablation studies to investigate the impact of architectural parameters, the effectiveness of fine-tuning, search space efficiency, and the discriminative performance of generated architectures. These studies reveal that larger filter sizes and specific numbers of stacks or modules enhance performance. Fine-tuning existing architectures can achieve nearly optimal results without separating NAS for each dataset. Furthermore, we analyze search space efficiency, uncovering patterns in frequently selected operations and architectural choices. This study highlights the strengths and efficiencies of PBC-NAS and BioNAS, providing valuable insights and guidance for future research and practical applications in biomedical image classification. [Display omitted] • PBC-NAS and BioNAS outperform state-of-the-art methods, AutoML and most of NAS studies. • BioNAS tops average accuracy; PBC-NAS leads in computational efficiency. • Fine-tuning existing architectures shows near-optimal results without separate NAS. • Larger filter sizes and specific stack/module configs lead to improved performance. • Analysis reveals patterns in operation selection and architectural choices for NAS. [ABSTRACT FROM AUTHOR]

Healthy eating is an important issue affecting a large part of the world population, so human diets are becoming increasingly popular, especially with the devastating consequences of Coronavirus Disease (Covid-19). A realistic and sustainable diet plan can help us to have a healthy eating habit since it considers most of the expectations from a diet without any restriction. In this study, the classical diet problem has been extended in terms of modelling, data sets and solution approach. Inspired by animals’ hunting strategies, it was re-modelled as a many-objective optimisation problem. In order to have realistic and applicable diet plans, cooked dishes are used. A well-known many-objective evolutionary algorithm is used to solve the diet problem. Results show that our approach can optimise specialised daily menus for different user types, depending on their preferences, age, gender and body index. Our approach can be easily adapted for users with health issues by adding new constraints and objectives. Our approach can be used individually or by dietitians as a decision support mechanism. [ABSTRACT FROM AUTHOR]

Airspace surveillance is a significant issue for many countries to control and manage their airspace. The number of radars used and their coverage rate are the main issues to consider in this case. Therefore, this paper addresses the problem of finding the best radar locations to obtain the highest coverage rate with the least possible number of radars in a certain region. The radar placement problem is considered as a multi-objective optimization problem with two objectives: the number of radars and the coverage rate. To perfectly solve this optimization problem, a set of multi-objective meta-heuristic approaches based on simulated annealing, memory-based steady-state genetic algorithm, a decomposition-based multi-objective algorithm with differential evolution, and non-dominated sorting genetic algorithm (NSGA-II) are utilized. Algorithms are tested on a dataset created using DTED-1 map elevation data for two different selected regions. Based on the results, the NSGA-II algorithm achieves the best results and the highest coverage ratios among the tested algorithms. Two improved versions of the NSGA-II algorithm are also proposed to enhance its performance and make it more suitable for solving this optimization problem. The experimental results show that a coverage rate of 98% could be achieved with a small number of radars, and by increasing the number of radars, it exceeds 99%. • This paper solves the radar placement problem using a set of metaheuristic approaches. • The problem is considered as a multi-objective optimization problem. • Two adaptive and enhanced versions of the NSGA-II algorithm are proposed. • The results show that a coverage rate of 98% can be achieved with just a few radars. [ABSTRACT FROM AUTHOR]