This paper proposes the application of the recently proposed hybrid particle swarm optimization (PSO) and firefly algorithm (FA), called HFPSO, for solving the mesh routers placement problem in wireless mesh networks (WMNs). HFPSO combines the local search ability of FA and the fast convergence ability of PSO algorithm. The effectiveness of HFPSO was demonstrated using many generated instances in comparison with FA and PSO algorithms taking into account the metrics of user coverage and network connectivity. The results showed that HFPSO is more effective than FA and PSO in finding optimal mesh routers locations.
The recent advances in online machine learning have open scope to experiment with different types of multimedia data. This paper has been presented in this direction using a popular second-order method as Newton step that guarantees a regret bound. We have used first- and second-order matrices of dimensions m (classes) by d (features) with pseudo-inverse technique and elementwise multiplication. This first- and second-order matrices size results in less storage and faster computations. The extensive experimentation in this paper has been performed on benchmarked datasets, where mistake rate, number of updates, and computation time are evaluated. The result indicates that our method outcome is at-par with state-of-the-art algorithms.
In this paper, a robust swarm-inspired algorithm has been proposed known as Cosine algorithm (CA) to solve the optimisation problem. The CA generates several initial random agents' solution and requires all of them to change towards or outwards the ideal solution by means of mathematical model on Cosine function. A number of adaptive and random variables are also added into this method to promote exploitation and exploration of the search space at certain optimization milestones. The results of performance metrics and test functions demonstrate that the developed algorithm is capable of successfully exploring diverse areas of a search space, avoiding local optima, converging towards the worldwide optimum and exploiting potential parts of a search space through optimisation.
In this study, low-velocity impact on fiber metal laminate (FML) plate with nanoclay reinforcement particles has been investigated. Rectangular FML specimens have been made by hand lay-up method using two 2024 aluminum sheets and 4 layers of basalt fibers. The ultrasonic device is used for nanoclay in samples, and the specimens are constructed in different weight percentages of nanoclay particles from 0 to 5%. Low-velocity impact test has been conducted using drop weight device at different level of impact energies. Artificial neural networks are designed and used for the prediction of the response factor (i.e., absorbed impact energy), while input factors are weight percentage of nanoclay particles and applied level of impact energy. The obtained experimental and numerical results showed that nanoclay reinforcement particles in specimens containing different weight percentages of nanoclay possess more contact force and energy absorption with respect to the non-nanoclay specimens.
Meta-heuristics have been widely used in both science and industry as reliable alternatives to conventional optimization algorithms to solve challenging, real-world problems. Despite being general-purpose and having a black-box nature, they require changes to solve multi-objective optimization problems. This paper proposes a multi-objective version of harmony search based on the archive. Archive, grid, and leader selection mechanisms are applied in multi-objectives of HS. Five real-engineering problems are evaluated with the results of three indexes. Based on the results, the AMHS is capable of providing acceptable results than other alternatives.
