Enhancing Speaker Recognition Models with Noise-Resilient Feature Optimization Strategies

Acoustics, Journal Year: 2024, Volume and Issue: 6(2), P. 439 - 469

Published: May 14, 2024

This paper delves into an in-depth exploration of speaker recognition methodologies, with a primary focus on three pivotal approaches: feature-level fusion, dimension reduction employing principal component analysis (PCA) and independent (ICA), feature optimization through genetic algorithm (GA) the marine predator (MPA). study conducts comprehensive experiments across diverse speech datasets characterized by varying noise levels counts. Impressively, research yields exceptional results different classifiers. For instance, TIMIT babble dataset (120 speakers), fusion achieves remarkable identification accuracy 92.7%, while various techniques combined K nearest neighbor (KNN) linear discriminant (LD) classifiers result in verification equal error rate (SV EER) 0.7%. Notably, this 93.5% SV EER 0.13% (630 speakers) using KNN classifier optimization. On white 630 accuracies 93.3% 83.5%, along values 0.58% 0.13%, respectively, were attained utilizing PCA (PCA-MPA) Furthermore, voxceleb1 dataset, PCA-MPA 95.2% 1.8%. These findings underscore significant enhancement computational speed performance facilitated strategies.

Language: Английский

---

Adaptive Marine Predator Optimization Algorithm (AOMA)-Deep Supervised Learning Classification (DSLC) Based IDS Framework for MANET Security

Intelligent and Converged Networks, Journal Year: 2024, Volume and Issue: 5(1), P. 1 - 18

Published: March 1, 2024

Due to the dynamic nature and node mobility, assuring security of Mobile Ad-hoc Networks (MANET) is one difficult challenging tasks today.In MANET, Intrusion Detection System (IDS) crucial because it aids in identification detection malicious attacks that impair network's regular operation.Different machine learning deep methodologies are used for this purpose conventional works ensure increased MANET.However, still has significant flaws, including algorithmic complexity, lower system performance, a higher rate misclassification.Therefore, goal paper create an intelligent IDS framework significantly enhancing MANET through use models.Here, minmax normalization model applied preprocess given cyber-attack datasets normalizing attributes or fields, which increases overall intrusion performance classifier.Then, novel Adaptive Marine Predator Optimization Algorithm (AOMA) implemented choose optimal features improving speed classifier.Moreover, Deep Supervise Learning Classification (DSLC) mechanism utilized predict categorize type based on proper training operations.During evaluation, results proposed AOMA-DSLC methodology validated compared using various measures benchmarking datasets.

Language: Английский

---

Cross-Coupled Dynamics and MPA-Optimized Robust MIMO Control for a Compact Unmanned Underwater Vehicle

Journal of Marine Science and Engineering, Journal Year: 2023, Volume and Issue: 11(7), P. 1411 - 1411

Published: July 14, 2023

A compact, 3-degrees-of-freedom (DoF), low-cost, remotely operated unmanned underwater vehicle (UUV), or MicroROV, is custom-designed, developed, instrumented, and interfaced with a PC for real-time data acquisition control. The nonlinear equations of motion (EoM) are developed the under-actuated, open-frame, cross-coupled MicroROV utilizing Newton-Euler approach. cross-coupling between heave yaw motion, an important dynamic class compact ROVs that barely reported, investigated here. This work thus motivated towards developing understanding physics highly coupled ROV model-based stabilizing controllers. linearized EoM aids in high-fidelity experimental data-driven transfer function models. heave-yaw model improved to auto-regressive moving average exogenous input (ARMAX) structure. acquired models facilitate use multi-parameter root-locus (MPRL) technique design baseline controllers multi-input, multi-output (MIMO) MicroROV. controller gains further optimized by employing innovative Marine Predator Algorithm (MPA). robustness designed gauged using gain phase margins. In addition, were deployed on onboard embedded system Simulink′s automatic C++ code generation capabilities. Finally, pool tests demonstrate efficacy proposed control strategy.

Language: Английский

---

Adaptive crossover-based marine predators algorithm for global optimization problems

Journal of Computational Design and Engineering, Journal Year: 2024, Volume and Issue: 11(4), P. 124 - 150

Published: June 26, 2024

Abstract The Marine Predators Algorithm (MPA) is a swarm intelligence algorithm developed based on the foraging behavior of ocean’s predators. This has drawbacks including, insufficient population diversity, leading to trapping in local optima and poor convergence. To mitigate these drawbacks, this paper introduces an enhanced MPA Adaptive Sampling with Maximin Distance Criterion (AM) horizontal vertical crossover operators – i.e., Crossover-based (AC-MPA). AM approach used generate diverse well-distributed candidate solutions. Whereas maintain diversity during search process. performance AC-MPA was tested using 51 benchmark functions from CEC2017, CEC2020, CEC2022, varying degrees dimensionality, findings are compared those its basic version, variants, numerous well-established metaheuristics. Additionally, 11 engineering optimization problems were utilized verify capabilities handling real-world problems. clearly show that performs well terms solution accuracy, convergence, robustness. Furthermore, proposed demonstrates considerable advantages solving problems, proving effectiveness adaptability.

Language: Английский

---

A comprehensive survey of honey badger optimization algorithm and meta-analysis of its variants and applications

Franklin Open, Journal Year: 2024, Volume and Issue: 8, P. 100141 - 100141

Published: Aug. 10, 2024

Metaheuristic algorithms are commonly used in solving complex and NP-hard optimization problems various fields. These have become popular because of their ability to explore exploit solutions problem domains. Honey Badger Algorithm (HBA) is a population-based metaheuristic algorithm inspired by the dynamic hunting strategy honey badgers, utilizing digging-seeking techniques. Since its introduction 2020, HBA has garnered widespread attention been applied across This review aims comprehensively survey improvement application problems. Additionally, conducts meta-analysis HBA's improvements, hybridization since introduction. According result survey, 52 studies presented improved using chaotic maps, levy flight mechanism, adaptive mechanisms, transfer functions, multi-objective mechanism opposition based learning techniques, 20 hybrid with other metaheuristics 101 uses original for wide acceptance within research community stems from straightforwardness, ease use, efficient computational time, accelerated convergence speed, high efficacy, capability address different kind issues, distinguishing it well-known approches presented.

Language: Английский

---

ICSOMPA: A novel improved hybrid algorithm for global optimisation

Evolutionary Intelligence, Journal Year: 2024, Volume and Issue: 17(5-6), P. 3337 - 3440

Published: May 8, 2024

Language: Английский

---

Expert knowledge data-driven based actor–critic reinforcement learning framework to solve computationally expensive unit commitment problems with uncertain wind energy

International Journal of Electrical Power & Energy Systems, Journal Year: 2024, Volume and Issue: 159, P. 110033 - 110033

Published: May 17, 2024

With the expansion of power grid, unaffordable computational cost and time will pose serious challenges time-efficient scheduling in unit commitment problem (UCP). However, existing optimization methods, i.e., mathematical programming methods meta-heuristic algorithms, are powerless time-consuming to handle computationally expensive UCP (CEUCP). Thus, reinforcement learning with strong inference time-saving performances motivated solve tackling CEUCPs. In this paper, a novel expert knowledge data-driven based actor–critic (AC) methodology is proposed for solving Specifically, AC methodology, knowledge, surrogate model, improved algorithm integrated further performance enhancement. Firstly, action selection mechanism (based on thermal units characteristic) into improve efficiency network training. Secondly, an extreme machine (ELM) model build reward function AC. detail, original replaced by lightweight ELM model. Shape distance enhancing accuracy. Finally, marine predators (MPA) obtaining optimal dispatching decisions rewards method quickly correctly. Original search pattern quantum representation boosting convergence. The excellent framework verified simulations 10-units, 100-units, 100-units wind energy test systems.

Language: Английский

---

Hybrid neural network wind speed prediction based on two-level decomposition and weighted averaging

Earth Science Informatics, Journal Year: 2024, Volume and Issue: 17(5), P. 4213 - 4232

Published: June 28, 2024

Language: Английский

---

Discrete Marine Predators Algorithm for Symmetric Travelling Salesman Problem

Evolutionary Intelligence, Journal Year: 2024, Volume and Issue: 17(5-6), P. 3833 - 3848

Published: July 2, 2024

Language: Английский

---

On Performance of Marine Predators Algorithm in Training of Feed-Forward Neural Network for Identification of Nonlinear Systems

Symmetry, Journal Year: 2023, Volume and Issue: 15(8), P. 1610 - 1610

Published: Aug. 20, 2023

Artificial neural networks (ANNs) are used to solve many problems, such as modeling, identification, prediction, and classification. The success of ANN is directly related the training process. Meta-heuristic algorithms extensively for training. Within scope this study, a feed-forward artificial network (FFNN) trained using marine predators algorithm (MPA), one current meta-heuristic algorithms. Namely, study aimed evaluate performance MPA in detail. Identification/modeling nonlinear systems chosen problem. Six applications. Some them static, some dynamic. Mean squared error (MSE) utilized metric. Effective testing results were obtained MPA. best mean values six 2.3 × 10−4, 1.8 10−3, 1.0 1.2 10−5, 2.5 10−4. compared with 16 have shown that better than other identification systems.

Language: Английский

---