Search

Bayesian optimization (BO) is commonly used to optimize the hyperparameters of transfer learning models to improve the model's performance significantly. In BO, the acquisition functions direct the hyperparameter space exploration during the optimization. However, the computational cost of evaluating the acquisition function and updating the surrogate model can become prohibitively expensive due to increasing dimensionality, making it more challenging to achieve the global optimum, particularly in image classification tasks. Therefore, this study investigates and analyses the effect of incorporating metaheuristic methods into BO to improve the performance of acquisition functions in transfer learning. By incorporating four different metaheuristic methods, namely Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC) Optimization, Harris Hawks Optimization, and Sailfish Optimization (SFO), the performance of acquisition function, Expected Improvement (EI), was observed in the VGGNet models for visual field defect multi-class classification. Other than EI, comparative observations were also conducted using different acquisition functions, such as Probability Improvement (PI), Upper Confidence Bound (UCB), and Lower Confidence Bound (LCB). The analysis demonstrates that SFO significantly enhanced BO optimization by increasing mean accuracy by 9.6% for VGG-16 and 27.54% for VGG-19. As a result, the best validation accuracy obtained for VGG-16 and VGG-19 is 98.6% and 98.34%, respectively. [ABSTRACT FROM AUTHOR]

We investigate the problem of scheduling a sequence of cars to be placed on an assembly line. Stations, along the assembly line install options (e.g. air conditioning), but have limited capacities, and hence cars requiring the same options need to be distributed far enough apart. The desired separation is not always feasible, leading to an optimisation problem that minimises the violation of the ideal separation requirements. In order to solve the problem, we use a large neighbourhood search (LNS) based on mixed integer programming (MIP). The search is implemented as a sliding window, by selecting overlapping subsequences of manageable sizes, which can be solved efficiently. Our experiments show that, with LNS, substantial improvements in solution quality can be found. [ABSTRACT FROM AUTHOR]

Due to different users' requirements, contemporary software has become feature-rich in terms of input functions (i.e., parameters) and selections (i.e., values). Exhaustive testing on sophisticated software systems is impractical as far as testing time and cost are concerned. Various test case design strategies have been proposed in the literature, such as equivalence class partitioning, boundary value analysis and decision tables. Unlike earlier works, combinatorial t-way testing supports the detection of faults caused by two or more input parameter interactions and thus efficiently minimizes the size of the test suite. Over the past few years, metaheuristic algorithms have appeared to be the most common choice for researchers since their effectiveness proves to offer optimal/near-optimal results. However, generating a t-way test suite is an NP-hard problem, and no single t-way strategy can guarantee to show superiority to others for all types of system configurations. Hence, this paper presents a new t-way strategy based on the Gravitational Search Algorithm (GSA), known as the Gravitational Search Test Generator (GSTG). The primary contribution of this paper is that GSA has adapted for the first time to t-way test data generation. The benchmarking results showcase that GSTG obtains competitive results in most system configurations compared to other existing strategies and addresses higher combination coverage (i.e., t ≤ 10). [ABSTRACT FROM AUTHOR]

Most EEG–based motor imagery classification research focuses on the feature extraction phase of machine learning, neglecting the crucial part for accurate classification which is the classification. In contrast, this paper concentrates on the classifier development where it thoroughly studies the performance analysis of k-Nearest Neighbour (k-NN) classifier on EEG data. In the literature, the Euclidean distance metric is routinely applied for EEG data classification. However, no thorough study has been conducted to evaluate the effect of other distance metrics to the classification accuracy. Therefore, this paper studies the effectiveness of five distance metrics of k-NN: Manhattan, Euclidean, Minkowski, Chebychev and Hamming. The experiment shows that the distance computations that provides the highest classification accuracy is the Minkowski distance with 70.08%. Hence, this demonstrates the significant effect of distance metrics to the k-NN accuracy where the Minknowski distance gives higher accuracy compared to the Euclidean. Our result also shows that the accuracy of k-NN is comparable to Support Vector Machine (SVM) with lower complexity for EEG classification.

Numerous research have demonstrated that Convolutional Neural Network (CNN) models are capable of classifying visual field (VF) defects with great accuracy. In this study, we evaluated the performance of different pre-trained models (VGG-Net, MobileNet, ResNet, and DenseNet) in classifying VF defects and produced a comprehensive comparative analysis to compare the performance of different CNN models before and after hyperparameter tuning and fine-tuning. Using 32 batch sizes, 50 epochs, and ADAM as the optimizer to optimize weight, bias, and learning rate, VGG-16 obtained the highest accuracy of 97.63 percent, according to experimental findings. Subsequently, Bayesian optimization was utilized to execute automated hyperparameter tuning and automated fine-tuning layers of the pre-trained models to determine the optimal hyperparameter and fine-tuning layer for classifying many VF defect with the highest accuracy. We found that the combination of different hyperparameters and fine-tuning of the pre-trained models significantly impact the performance of deep learning models for this classification task. In addition, we also discovered that the automated selection of optimal hyperparameters and fine-tuning by Bayesian has significantly enhanced the performance of the pre-trained models. The results observed the best performance for the DenseNet-121 model with a validation accuracy of 98.46% and a test accuracy of 99.57% for the tested datasets. [ABSTRACT FROM AUTHOR]

This paper aims to tackle the Patient Admission Scheduling Problem (PASP) using the Discrete Flower Pollination Algorithm (DFPA), a new, meta-heuristic optimization method based on plant pollination. PASP is one of the most important problems in the field of health care. It is a highly constrained and combinatorial optimization problem of assigning patients to medical resources in a hospital, subject to predefined constraints, while maximizing patient comfort. While the flower pollination algorithm was designed for continuous optimization domains, a discretization of the algorithm has been carried out for application to the PASP. Various neighborhood structures have been employed to enhance the method, and to explore more solutions in the search space. The proposed method has been tested on six instances of benchmark datasets for comparison against another algorithm using the same dataset. The prospective method is shown to be very efficient in solving any scheduling problem., (Copyright © 2021 Elsevier Ltd. All rights reserved.)