This paper introduces a novel Nature-inspired Multi-Objective Particle Swarm Optimization algorithm for Multifactorial Problems, called MOPSO-mfact. The MOPSO-mfact algorithm operates in two processing steps: an initial pre-search in a unified search space, followed using a dominance operator to identify the best skill factor. Subsequently, individuals with identical skill factors are grouped into sub-swarms for multitasking optimization, avoiding random mating probabilities. MOPSO-mfact was evaluated using 36 Multifactorial (ETMOF) benchmark suite. Comparative analysis using the Inverted General Distance (IGD) and Mean Inverted General Distance (MIGD) metrics, along with the Mean Standard Score (MSS), determined the best multitasking approach. Parameter optimization was achieved via sensitivity analysis with the Taguchi method. MOPSO-mfact demonstrated promising results, achieving a strong MSS for 28 ETMOF problems and solving 33 out of 36 problems based on MIGD metrics.

Background With the rapid development of artificial intelligence technology, highly intelligent and unmanned factories have become an important trend. In the complex environments of smart factories, the long-term tracking and inspection of specified targets such as operators and special goods, as well as comprehensive visual recognition and decision-making capabilities throughout the whole production process, are critical components of automated unmanned factories. It is inevitable that issues such as target occlusion and disappearance will occur, thus exacerbating the long-term tracking challenge. Currently, there are no long-term tracking studies specifically addressing occlusions in these environments. Methods We first construct three new benchmark datasets in the complex workshop environment of a smart factory (referred to as SF-Complex3 data), which include challenging conditions such as complete occlusion and partial occlusion of targets. Next, utilising a brain memory inspired approach, we determine uncertainty estimation parameters: confidence, peak-to-sidelobe ratio (PSR), and average peak to correlation energy (APCE), to derive a continual learning based adaptive model update method. Additionally, we design a lightweight target detection model to automatically detect and locate targets in the initial frame and during re-detection. Finally, we integrate the algorithm with ground mobile robots and UAV-based imaging and processing equipment to build a new visual detection and tracking framework, termed SFC-RT (Smart Factory Complex Tracking and Identification). Results We conducted extensive tests on the benchmark UAV20L and SF-Complex3 datasets. Compared to state-of-the-art tracking algorithms, our proposed algorithm demonstrates an average performance improvement of 6% when facing key challenging attributes. Moreover, it can smoothly run on embedded platforms, including mobile robots and UAVs, at a real-time speed of 36.4 fps. Conclusions The proposed SFC-RT framework is shown to efficiently and accurately address the challenges of target loss and occlusion for long-term tracking within complex smart factory environments. It meets the requirements of real-time performance, robustness, and lightweight design.

This paper presents a new Multi-Objective Particle Swarm Optimization algorithm for Multifactorial Optimization Problems referred to as MOPSO-mfact. It operates in two distinct steps with two processing steps. Initially, a pre-search is conducted in a unified search space where all individuals are optimized to address various tasks simultaneously. Then, the dominance operator is used to explore the solution encoding search space and identify the best skill factor. In the subsequent step, individuals with identical solution skill factors are grouped together to create multiple sub-swarms for multitasking optimization. This approach leverages the dominance operator instead of relying on random mating probabilities. The MOPSO-mfact algorithm was tested on a set of Multifactorial test benchmarks called “ETMOF” which includes 36 multi/many objectives optimization problems. A comparative study is conducted toward the Inverted General Distance (IGD) and the Mean Inverted General Distance (MIGD) metrics. The Mean Standard Score (MSS) is used to determine the best approach for multitasking optimization. Parameter optimization for MOPSO-mfact is achieved through sensitivity analysis using the Taguchi method, ensuring optimal performance. The MOPSO-mfact algorithm demonstrated promising results, achieving a good MSS result for solving the 28/26 ETMOF test suite as assessed by the IGD indicators. Additionally, it performed well, solving 33 out of 36 problems in the ETMOF suite when evaluated using the MIGD metric.

sion dataset. Results: By amalgamating features extracted from all three CNNs and utilizing the medium Gaussian kernel of the SVM classifier, our proposed system achieves an outstanding av- erage classification accuracy of 90.4%. Conclusions: Our developed MPXCN-Net is suitable for test- ing with a large diversified dataset before being used in clin- ical settings.

Particle swarm optimization system based on the distributed architecture over multiple sub-swarms has shown its efficiency for static optimization and has not been studied to solve dynamic multi-objective problems (DMOPs). When solving DMOP, tracking the best solutions over time and ensuring good exploitation and exploration are the main challenging tasks. This study proposes a novel Dynamic Pareto bi-level Multi-Objective Particle Swarm Optimization (DPb-MOPSO) algorithm including two parallel optimization levels. At the first level, all solutions are managed in a single search space. When a dynamic change is successfully detected in the objective values, the Pareto ranking operator is used to enable a multiple sub-swarm’ subdivisions and processing which drives the second level of enhanced exploitation. A dynamic handling strategy based on random detectors is used to track the changes of the objective function due to time-varying parameters. A response strategy consisting in re-evaluate all unimproved solutions and replacing them with newly generated ones is also implemented. Inverted generational distance, mean inverted generational distance, and hypervolume difference metrics are used to assess the DPb-MOPSO performances. All quantitative results are analyzed using Friedman’s analysis of variance while the Lyapunov theorem is used for stability analysis. Compared with several multi-objective evolutionary algorithms, the DPb-MOPSO is robust in solving 21 complex problems over a range of changes in both the Pareto optimal set and Pareto optimal front. For 13 UDF and ZJZ functions, DPb-MOPSO can solve 8/13 and 7/13 on IGD and HVD with moderate changes. For the 8 FDA and dMOP benchmarks, DPb-MOPSO was able to resolve 4/8 with severe change on MIGD, and 5/8 for moderate and slight changes. However, for the 3 kind of environmental changes, DPb-MOPSO assumes 4/8 of the solving function on IGD and HVD metrics.

Fake account detection is a topical issue when many Online Social Networks encounter several issues caused by the growing number of unethical online activities. This study presents a new Quantum Beta-behaved Multi-Objective Particle Swarm Optimization (QB-MOPSO) algorithm for machine learning based Twitter fake accounts detection. The proposed approach aims to improve the learning process of deep neural networks, random forest, through minimizing simultaneously the feature dimensionality and the classification error rate. The main contribution consists in proposing a quantum beta MOPSO to handle the training phase of neural and deep architectures. The QB-MOPSO is used to perform a multi-objective training of the random forest algorithm. The QB-MOPSO has two optimization profiles: the first one uses a quantum-behaved equation for improving the exploratory behaviour of PSO, while the second one uses a beta function to enhance PSO’s exploitation. An extensive experimental study is carried out using two open Twitter datasets with 1982 and 928 accounts. The new proposal is a random forest QB-MOPSO. Results showed that random forest QB-MOPSO accuracy is about 99.19% and 97.52% accounts on datasets 1 and 2. Comparative analysis of the prosed architecture toward the original architecture showed that the use of QB-MOPSO for learning enhances the random forest algorithm which perform then the original ones.

Sign language is a mean of communication between the deaf community and hearing people, who use hand gestures, facial expressions, and body language to communicate. It has the same level of complexity as spoken language, but it does not employ the same sentence structure as English. The motions in sign language are made up of a range of distinct hand and finger articulations that are occasionally synchronized with the head, face, and body. Existing sign language recognition systems are mainly camera-based, which have fundamental limitations of poor lighting conditions, potential training challenges with longer video sequence data, and serious privacy concerns. This study presents a contact-less and privacy-preserving British sign language (BSL) Recognition system using Radar and deep learning algorithms, namely Inceptionv3, VGG16, and VGG19. The six most common emotions are considered, namely confused, depressed, happy, hate, lonely, and sad. The collected data is represented in the form of spectrograms. The deep learning models, InceptionV3, VGG19, and VGG16 then extract spatiotemporal features from the Spectrogram. Finally, the BSL emotions are accurately identified by classifying the Spectrograms, into the considered emotions signs. The simulation results demonstrate that a maximum classifying accuracy of 93.33% is obtained using VGG16.

Particle swarm optimization system based on the distributed architecture has shown its efficiency for static optimization and has not been studied to solve dynamic multiobjective problems (DMOPs). When solving DMOP, tracking the best solutions over time and ensuring good exploitation and exploration are the main challenging tasks. This study proposes a novel Dynamic Pareto bi-level Multi-Objective Particle Swarm Optimization (DPb-MOPSO) algorithm including two parallel optimization levels. At the first level, all solutions are managed in a single search space. When a dynamic change is successfully detected, the Pareto ranking operator is used to enable a multiswarm subdivisions and processing which drives the second level of enhanced exploitation. A dynamic handling strategy based on random detectors is used to track the changes of the objective function due to time-varying parameters. A response strategy consisting in re-evaluate all unimproved solutions and replacing them with newly generated ones is also implemented. Inverted generational distance, mean inverted generational distance, and hypervolume difference metrics are used to assess the DPb-MOPSO performances. All quantitative results are analyzed using Friedman’s analysis while the Lyapunov theorem is used for stability analysis. Compared with several multi-objective evolutionary algorithms, the DPb-MOPSO is robust in solving 21 complex problems over a range of changes in both the Pareto optimal set and Pareto optimal front. For 13 UDF and ZJZ functions, DPb-MOPSO can solve 8/13 and 7/13 on IGD and HVD with moderate changes. For the 8 FDA and dMOP benchmarks, DPb-MOPSO was able to resolve 4/8 with severe change on MIGD, and 5/8 for moderate and slight changes. However, for the 3 kind of environmental changes, DPb-MOPSO assumes 4/8 of the solving function on IGD and HVD.

Online Social Networks (OSNs) have grown exponentially in the last few years due to their applications in real life like marketing, recommendation systems, and social awareness campaigns. One of the most important research areas in this field is Influence Maximization (IM). IM pertains to finding methods to maximize the spread of information (or influence) across a social network. Previous works in IM have focused on using a pre-defined edge propagation probability or using the Hurst exponent (H) to identify which nodes to be activated. This is calculated on the basis of self-similarity in the time series depicting a user’s (node) past temporal interaction behaviour. In this work, we propose a Time Series Characteristic based Hurst-based Diffusion Model (TSC-HDM). The model calculates Hurst Exponent (H) based on the stationary or non-stationary characteristic of the time series. Furthermore, our model selects a handful of seed nodes and activates every seed node’s inactive successor only if H>0.5 . The process is continued until the activation of successor nodes is not possible. The proposed model was tested on 4 datasets - UC Irvine messages, Email EU-Core, Math Overflow 3, and Linux Kernel mailing list. We have also compared the results against 4 other Influence Maximisation models - Independent Cascade (IC), Weighted Cascade (WC), Trivalency (TV), and Hurst-based Influence Maximisation (HBIM). Our model achieves as much as 590% higher expected influence spread as compared to the other models. Moreover, our model attained 344% better average influence spread than other state-of-the-art models.

The prediction of news popularity is having substantial importance for the digital advertisement community in terms of selecting and engaging users. Traditional approaches are based on empirical data collected through surveys and applied statistical measures to prove a hypothesis. However, predicting news popularity based on statistical measures applied to past data is highly questionable. Therefore, in this paper, we predict news popularity using machine learning classification models and deep residual neural network models. Articles are usually made up of textual content and in many cases, images are also used. Although it is evident that the appropriate amount of textual data is required to extract features and create models, image data is also helpful in gaining useful information. In this paper, we present a novel multimodal online news popularity prediction model based on ensemble learning. This research work acts as a guide for extensive feature engineering, feature extraction, feature selection, and effective modeling to create a robust news popularity Prediction Model. Three kinds of features – meta features, text features, and image features are used to design an influential and robust model. The performance measure Root Mean Squared logarithmic error (RMSLE) is used to validate the outcome of the proposed model. Further, the most important features are sought out for the proposed model to verify the dependence of the model on text and image features.