A benchmark dataset for multi-objective flexible job shop cell scheduling.

This data article presents a description of a benchmark dataset for the multi-objective flexible job shop scheduling problem in a cellular manufacturing environment. This problem considers intercellular moves, exceptional parts, sequence-dependent family setup and intercellular transportation times, and recirculation requiring minimization of makespan and total tardiness simultaneously. It is called a flexible job shop cell scheduling problem with sequence-dependent family setup times and intercellular transportation times (FJCS-SDFSTs-ITTs) problem. The dataset has been developed to evaluate the multi-objective evolutionary algorithms of the FJCS-SDFSTs-ITTs problems that are presented in 'Evolutionary algorithms for multi-objective flexible job shop cell scheduling'. The dataset contains forty- three benchmark instances from 'small' to 'large', including a large real-world problem instance. Researchers can use the dataset to evaluate the future algorithms for the FJCS-SDFSTs- ITTs problems and compare the performance with the existing algorithms.

Multi-objective topology optimisation for acoustic porous materials using gradient-based, gradient-free, and hybrid strategies.

When designing passive sound-attenuation structures, one of the challenging problems that arise is optimally distributing acoustic porous materials within a design region so as to maximise sound absorption while minimising material usage. To identify efficient optimisation strategies for this multi-objective problem, several gradient, non-gradient, and hybrid topology optimisation strategies are compared. For gradient approaches, the solid-isotropic-material-with-penalisation method and a gradient-based constructive heuristic are considered. For gradient-free approaches, hill climbing with a weighted-sum scalarisation and a non-dominated sorting genetic algorithm-II are considered. Optimisation trials are conducted on seven benchmark problems involving rectangular design domains in impedance tubes subject to normal-incidence sound loads. The results indicate that while gradient methods can provide quick convergence with high-quality solutions, often gradient-free strategies are able to find improvements in specific regions of the Pareto front. Two hybrid approaches are proposed, combining a gradient method for initiation and a non-gradient method for local improvements. An effective Pareto-slope-based weighted-sum hill climbing is introduced for local improvement. Results reveal that for a given computational budget, the hybrid methods can consistently outperform the parent gradient or non-gradient method.

Identifying Chirality in Line Drawings of Molecules Using Imbalanced Dataset Sampler for a Multilabel Classification Task.

Chirality, the ability of some molecules to exist as two non-superimposable mirror images, profoundly influences both chemistry and biology. Advances in deep learning enable the automatic recognition of chemical structure diagrams, however, studies on discovering the molecule chirality are scarce and the machine-readable molecular representations are not always sufficient to fully support the encoding of this important property. Here, we pretrained networks on a ChEMBL+ dataset (79641 molecules) and fine-tuned them for the binary classification of chirality (achiral/chiral) or multilabel chirality type classifications (none/centre/axial/planar). To address the label combination imbalanced problem in the multilabel task, the study proposed a Formulated Imbalanced Dataset Sampler (FIDS) to sample a formulated amount of minority label combinations on top of the training set. On a 10-fold cross validation experiment using our CHIRAL dataset (1142 manually curated molecules), our models achieved up to an accuracy of 90 % in the binary task. In the multilabel task incorporated with FIDS, the overall performance increases from 87 % to 89 % and the accuracy per label combination can attained up to a 50 % increase. Through the study of heatmaps, our work also exemplified the potential of deep neural network to make predictions based on the actual location of chirality elements.

Comparison of heuristics and metaheuristics for topology optimisation in acoustic porous materials.

When designing sound packages, often fully filling the available space with acoustic materials is not the most absorbing solution. Better solutions can be obtained by creating cavities of air pockets, but determining the most optimal shape and topology that maximises sound absorption is a computationally challenging task. Many recent topology optimisation applications in acoustics use heuristic methods such as solid-isotropic-material-with-penalisation (SIMP) to quickly find near-optimal solutions. This study investigates seven heuristic and metaheuristic optimisation approaches including SIMP applied to topology optimisation of acoustic porous materials for absorption maximisation. The approaches tested are hill climbing, constructive heuristics, SIMP, genetic algorithm, tabu search, covariance-matrix-adaptation evolution strategy (CMA-ES), and differential evolution. All the algorithms are tested on seven benchmark problems varying in material properties, target frequencies, and dimensions. The empirical results show that hill climbing, constructive heuristics, and a discrete variant of CMA-ES outperform the other algorithms in terms of the average quality of solutions over the different problem instances. Though gradient-based SIMP algorithms converge to local optima in some problem instances, they are computationally more efficient. One of the general lessons is that different strategies explore different regions of the search space producing unique sets of solutions.

A study on offshore wind farm siting criteria using a novel interval-valued fuzzy-rough based Delphi method.

This study investigates the degree of importance of criteria affecting the optimal site selection of offshore wind farms. Firstly, forty two different influential criteria have been selected by reviewing the scientific literature on offshore wind farm site selection. Secondly, a survey has been conducted receiving a response from thirty four internationally renowned experts across seventeen countries. Each participant is asked to indicate the importance and relevance of each criterion based on their experience. Finally, the importance of each criterion for offshore wind farm site selection is determined using a novel Decision Making-Level Based Weight Assessment (LBWA) approach based on interval-valued fuzzy-rough numbers (IVFRN). The proposed method allows exploitation of the uncertainties and subjectivity that exist in the decision-making process. The results from this study improve our understanding of the importance and impact of each criterion which we believe would be invaluable for the future studies on the site selection of offshore wind farms.

A Case Study of Controlling Crossover in a Selection Hyper-heuristic Framework Using the Multidimensional Knapsack Problem.

Hyper-heuristics are high-level methodologies for solving complex problems that operate on a search space of heuristics. In a selection hyper-heuristic framework, a heuristic is chosen from an existing set of low-level heuristics and applied to the current solution to produce a new solution at each point in the search. The use of crossover low-level heuristics is possible in an increasing number of general-purpose hyper-heuristic tools such as HyFlex and Hyperion. However, little work has been undertaken to assess how best to utilise it. Since a single-point search hyper-heuristic operates on a single candidate solution, and two candidate solutions are required for crossover, a mechanism is required to control the choice of the other solution. The frameworks we propose maintain a list of potential solutions for use in crossover. We investigate the use of such lists at two conceptual levels. First, crossover is controlled at the hyper-heuristic level where no problem-specific information is required. Second, it is controlled at the problem domain level where problem-specific information is used to produce good-quality solutions to use in crossover. A number of selection hyper-heuristics are compared using these frameworks over three benchmark libraries with varying properties for an NP-hard optimisation problem: the multidimensional 0-1 knapsack problem. It is shown that allowing crossover to be managed at the domain level outperforms managing crossover at the hyper-heuristic level in this problem domain.