We present TellTale, a text-only approach to ambivalence/hesitancy (A/H) recognition in interview videos, evaluated on the BAH dataset as part of the 3rd A/H Video Recognition Challenge (11th ABAW Workshop, ECCV 2026). Although the dataset provides video, audio, facial crops, and transcripts, TellTale relies on the transcript alone and combines three probability streams. Two text encoders, multilingual-e5-large and mDeBERTa-v3-base, are fine-tuned with parameter-efficient LoRA adapters under a multiple-instance learning (MIL) objective, in which transcript chunks are scored individually and pooled with a smooth maximum so that only the video-level label is needed for supervision. The third stream requires no training: a quantized 14B instruction LLM is prompted, zero-shot, to rate each transcript for A/H. The three probabilities are combined by a weighted average and a single decision threshold, both selected on participant-grouped cross-validated predictions. On the organizer-scored private test set of 152 videos from unseen participants, TellTale achieves a Macro-F1 of 0.7364 and an average precision of 0.7940, compared with 0.2827 Macro-F1 for the official vision-based baseline.

This paper presents the enzyme action optimization (EAO) algorithm, a novel bio-inspired optimization algorithm designed to simulate the adaptive enzyme mechanism in biological systems. EAO employs a novel strategy that dynamically balances between exploration and exploitation to efficiently navigate and optimize complex, multi-dimensional search spaces. EAO has been tested over diverse benchmark datasets, including the 23 classical benchmark functions, IEEE CEC2017, CEC2022 benchmark functions, where it has been compared with 14 recent and highly cited optimizers. The results show the superior performance of EAO over the compared optimizers in terms of finding the optimal solution, convergence speed, robustness, and overall performance. Furthermore, EAO was applied to solve five engineering design problems and demonstrated excellent performance results. The source code of EAO is publicly available for both MATLAB at: https://www.mathworks.com/matlabcentral/fileexchange/170296-enzyme-action-optimizer-a-novel-bio-inspired-optimization and PYTHON at: https://github.com/AliRodan/Enzyme-Action-Optimizer .

Stellar Oscillation Optimizer (SOO) takes its core inspiration from the study of stellar pulsations, a domain often referred to as asteroseismology which is formulated as an optimization algorithm for continuous domain. In this paper, the Binary version of Stellar Oscillation Optimizer (BSOO) is proposed for Feature Selection (FS) problems. BSOO introduces binary adaptations, including threshold-based encoding, controlled oscillatory movements, and a top-solution influence mechanism. In order to evaluate the BSOO, sixteen FS datasets are used with different numbers of features, samples, and class labels. Seven performance measures are also used, which are: fitness value, number of selected features, accuracy, sensitivity, specificity, Precision, and F-measure. An intensive comparative evaluation against 18 state-of-the-art optimization algorithms using the same datasets has been conducted. The results show that the proposed BSOO version is able to compete well with the other FS-based methods where it is able to overcome several methods and produce the best overall results for some datasets on different measurements. Furthermore, the convergence behavior to show the optimization behavior of BSOO during the search is investigated and visualized. Interestingly, the BSOO is able to provide a suitable trade-off between the global wide-range exploration and local nearby exploitation during the optimization process. This is proved using the statistical Wilcoxon Rank-Sum Test Results. In conclusion, this paper provides a new alternative solution for FS research community that is able to work well for many FS instances and find the optimal solution. The source code of BSOO is publicly available for both MATLAB at: https://www.mathworks.com/matlabcentral/fileexchange/180096-bsoo-binary-stellar-oscillation-optimizer and PYTHON at: https://github.com/AliRodan/BSOO-Binary-Stellar-Oscillation-Optimizer.

This paper introduces the Enhanced Frilled Lizard Optimizer (EFLO), an improved metaheuristic algorithm based on the recently proposed Frilled Lizard Optimizer. EFLO addresses key limitations, such as premature convergence and limited local exploitation, by incorporating four main enhancements: an adaptive step size sensitive to population diversity, a top-K prey selection mechanism guiding exploration, a dual-vantage tree-climbing strategy for effective exploitation, and a targeted Gaussian local search focused exclusively on optimal solutions. Extensive benchmarking on twelve single-objective IEEE CEC 2022 functions with dimensions (D = 10) and higher dimensional settings (D = 20) demonstrates EFLO's superior performance. EFLO achieves the lowest mean errors on eleven functions and consistently ranks first overall, significantly outperforming the original FLO and twenty-one recent and state-of-the-art optimizers. Statistical analyses, including the Wilcoxon signed-rank test, confirm EFLO's significant superiority when compared to benchmark optimizers and recent state-of-the-art algorithms. Practical applicability is further validated on five engineering design problems, where EFLO consistently yields optimal or near-optimal solutions with significantly reduced variability. Notably, EFLO achieves a 9.9% mass reduction in robot gripper design, a 1.0% improvement in spring weight optimization, and achieves solutions very close to theoretical or global optima in cantilever beam, three-bar truss, and welded beam problems. These results highlight EFLO's robustness and effectiveness in addressing complex, high-dimensional, and engineering design optimization tasks. The source code of EFLO is publicly available at: https://www.mathworks.com/matlabcentral/fileexchange/181623-enhanced-frilled-lizard-optimizer-eflo.

The widespread use of metaheuristic optimization algorithms across engineering, machine learning, and scientific modeling has led to significant advancements in solving complex, high-dimensional, and nonlinear problems. However, a persistent limitation in most existing metaheuristics is their inherent reliance on stochastic operators, random sampling, mutation, and probabilistic decisions. Although these stochastic elements enhance exploration capabilities, they compromise reproducibility and introduce variability in results across repeated runs. This inconsistency presents challenges in safety-critical and industrial applications where deterministic behavior and result traceability are essential. In this paper, we propose the Three-Body Deterministic Optimizer (TBD) to address these gaps. This novel metaheuristic eliminates all stochastic elements and introduces a fully deterministic search process inspired by orbital mechanics and chaos theory. Drawing on the classical three-body problem, TBD formulates two maneuver mechanisms, the Newtonian and Lagrangian maneuvers, to simulate gravitational attraction and multi-body interactions. These maneuvers are driven by a chaotic logistic map, which introduces complex yet fully reproducible dynamics. TBD is evaluated on the CEC2017 and CEC2022 benchmark suites, as well as on real-world engineering design problems such as cantilever beam, spring, and pressure vessel optimization, together with its application to Convolutional Neural Network (CNN) hyperparameter tuning. Experimental results show that TBD consistently achieves high-quality solutions, matching or outperforming state-of-the-art stochastic optimizers in accuracy and convergence speed, while maintaining strict determinism. The findings demonstrate that deterministic metaheuristics can retain the exploration-exploitation balance necessary for global optimization, offering a promising alternative for applications demanding transparency, repeatability, and computational rigor. The source code of TBD is publicly available for both MATLAB at: ( https://www.mathworks.com/matlabcentral/fileexchange/182063-three-body-deterministic-optimizer-tbd ) and PYTHON at: ( https://github.com/AliRodan/Three-Body-Deterministic-Optimizer-TBD ).

In this paper, we introduce an enhanced Particle Swarm Optimization using reflection strategy called Reflected Particle Swarm Optimization (RPSO) algorithm, RPSO designed to enhance robot path planning in complex environments such as static environment with obstacles. The integration aims to improve the exploration and exploitation capabilities of the search process. RPSO has been compared with state-of-art and well-known optimization algorithms. The results demonstrate that RPSO consistently outperforms the compared algorithms in finding the shortest paths and maintaining high levels of consistency across different trials. Furthermore, the results show that RPSO, with its innovative reflection mechanism, offers a significant advancement in robot path planning.

Semantic-enhanced recommendation systems are promising approaches to overcome the sparsity and cold-start problems, which are hard to handle using the conventional collaborative filtering (CF) approaches. Further research is needed to effectively integrate ontologies into collaborative filtering recommender systems. This paper proposes an ontology-based semantic similarity measure to evaluate similarities between items and eventually generate accurate recommendations. The proposed semantic similarity measure termed fusion-based semantic similarity takes into account the semantics of ontological instances (i.e. items) inferred from a specific domain ontology, which is determined by analyzing the hierarchical relationships among the instances, as well as the features of the instances and their relationships to other instances. The new measure comprehensively captures the semantic knowledge associated with instances by exploiting all possible shared semantics between instances in a given domain ontology. Furthermore, this paper proposes a new semantic-enhanced hybrid recommendation approach as a result of combining the new semantic similarity measure with the standard item-based CF to enhance the quality of generated recommendations. In order to assess the effectiveness of our semantic-enhanced hybrid collaborative filtering method, a series of experiments were conducted to compare the performance of the proposed approach against well-established benchmark techniques. The reported experimental results consistently emphasize its superiority, demonstrating enhanced predictive abilities and a notable improvement in the quality of recommendations. More specifically, the proposed approach achieved notable 6% reduction in Mean Absolute Error (MAE) in certain cases, outperforming other benchmark techniques. Additionally, this study highlights the potential of using semantic-based similarity to enhance the performance of recommendation systems. Such enhancements address challenges within collaborative filtering, potentially leading to advancements in recommendation system design and optimization.

Recurrent Neural Network (RNN) can offer more expressive power to approximate nonlinear dynamical systems compared to Feedforward Neural Network (FFNN). Their hidden layer output can be fed back into the network to serve as an additional input, which reserves the past information in the unit activation function. RNN is currently used in multitude of applications, including regression and classification challenges. Typically, they are trained using Backpropagation Through Time (BPTT) or Real-time Recurrent Learning (RTRL). In this paper, we propose the use of Negative Correlation Learning (NCL), an ensemble learning model for training Recurrent Neural Networks whose optimal weights are obtained using NCL with BPTT or NCL with RTRL. To the best of our knowledge, this is the first study that applies NCL style training for any state space modelling (e.g. RNNs ensembles). Experiments results based on four datasets confirms that NCL based RNNs ensemble approach achieves better generalization performance in comparison with flat RNNs ensemble (without NCL).

Clock synchronization in the Mac layer plays a vital role in wireless sensor network communication that maintains time-based channel sharing and offers a uniform timeframe among different network nodes. Most wireless sensor networks are distributed where no common clock exists among them. Therefore, joint actions are realized by exchanging messages, with time stamps using local sensor clocks. These clocks can easily drift seconds and cause functional problems to the applications that depend on time synchronization. Time synchronization is a major and challenging factor in wireless sensor networks that needs to be studied and explored. In this paper, we propose integrated time synchronization protocols that serve wireless sensor network applications under normal, secured, and unreliable environments. The proposed protocols are discussed and evaluated based on their accuracy, cost, hierarchy, reliability, and security. Simulation results show that the proposed time synchronization protocols outperform the state-of-the-art techniques in achieving a minimum synchronization time.

Sentiment Analysis is currently considered as one of the most attractive research topics in Natural Language Processing (NLP) field. The main objective of sentiment analysis is to identify the opinions and emotions of the users through written contents. While there are different studies that have approached this field using various techniques, it is still considered a challenging topic with many difficulties that are yet to be solved, such as having modern accents, slang words, spelling and grammatical mistakes, and other issues that cannot be overcome with traditional methods and sentiment lexicons. In this work, we propose a hybrid machine learning approach to enhance sentiment analysis; as we build a classification model based on three classes, which are positive, neutral, and negative emotions, using Support Vector Machines (SVM) classifier, while combining two feature selection techniques using the ReliefF and Multi-Verse Optimizer (MVO) algorithms. We also extract more than 6900 tweets from Twitter social network to test our work. Our hybrid method is compared against other classifiers and methods in terms of accuracy. Results show that our proposed method outperforms other techniques and classifiers, by obtaining better results in most of the datasets while reducing the number of features by up to 96.85% from the original feature set. We also categorize the extracted features into Objective, Subjective and Emoticon words to analyze them during the first and the final feature selection processes and find any existing relations. Very similar results are obtained by both feature selection techniques; due to a number of factors that are explained in this paper.

Due to the constant evolvement of the web and the viral spread of online news on social media, predicting the popularity of a news article became a topic of interest to many categories of people ranging from marketing personnel to politicians. In this paper, we focus on comparing four classification algorithms on a dataset consisting of 39000 news articles taken from Mashable website. The articles were classified into two classes: Popular and not popular. Four different machine learning algorithms were used for classification of the data (KNN, Naïve bayes, Adaboost, and decision tree). Finally, the four classification methods were compared with each other.

A grid is a system that can manage and organize services and resources that spread amongst different control domains, employ interfaces and protocols, and offer a high quality of services.The integration of Multi-Agent Systems (MAS) with a grid environment significantly affects grid performance.MAS is considered a suitable solution for open systems that modify frequently.Grid offers a wide range of resources for its users, and some of these resources might not be used or utilized for some time before any new jobs come to the grid for processing.Usually, these resources are accumulated in massive data centers to cater to the grid users' growing demand.This accumulated will lead to consuming a considerable amount of electricity for their operation.The ever-increasing usage of grid computing has led to an increase in electrical energy usage by massive servers in their data centers.In this paper, we have proposed an automated system composed of modern agents that can be used the power wastage resulted from inactive servers inside the data centers for a specific amount of time.The proposed technique depends on switching inactive virtual and physical machines to lower power positions (Sleep/Wakeup or switched off) while still preserving customers' performance requirements.The Automated system has been tested and evaluated using the Jade tools.The results show that the newly proposed method can reduce power wastage for inactive grid resources.

In this paper, a fuzzy logic controller will be introduced and discussed that can be applied for manufacturing systems. Our proposed control system will utilize fuzzy logic principles to find the optimum number of machines and operators assignment by developing the heuristic relations between the inputs and the outputs of the controller. The project inputs are the desired daily production and the average product processing time by each machine. The production process is assumed to be automated and the routing of products to these machines will be automated and requires negligible time. All machines are also assumed identical in both the functionality (manufacturing process) and the service time. Given the inputs as the number of products and the average service time, the goal of this controller will be to determine the number of the machines and the number of the operators needed to meet the production demand.

Grid Computing refers to systems and applications that incorporate and control distributed services and resources to resolve systematic or industrial issues. Over the past few years, developers have realized the need for an automatic mechanism that can be used in employing the grid power, improving its operating and enhancing its production. Multi-Agents Systems are generally employed to solve problems by using decentralized techniques by a set of agents collaborate to resolve a challenge. Hence, it is considered suitable solutions for open systems that change frequently. In this paper, we propose a modern architecture that utilizes a Multi Agent System not only to help in choosing the best resources but also to generate the negotiation protocol between the grid users and providers to fully deploy the capacity of grid computing. Moreover, the proposed architecture can play the main role in monitoring the list of users' jobs as they are being handled.

Security is nonnegotiable key point for in-vehicle networking. However, all communication between Electrical Control Unites (ECU) still suffer from security drawbacks like highly processing time or preserving confidentiality, integrity and authenticity. In this paper, we propose an approach to assess the feasibility of a private Blockchain technology to overcome the aforementioned drawbacks. In this approach, we consider in-vehicle networking contains two parts, namely, central (or connected) gateway (cGW) and switches. cGW and switches are Blockchain nodes, wherein Blockchain consensus protocols are what keep all the nodes on a network synchronized with each other. The approach considers any communication type between ECUs as an individual event, which can be a transaction, data entry or application execution. A use case of secure communication between two ECUs is presented as an evaluation mechanism for securing in-vehicle networking using the proposed Blockchain approach.

Echo state networks (ESN) are a special type of recurrent neural networks (RNN) with distinct performance in the field of reservoir computing (RC). The state space of the ESN is initially randomised and the reservoir weights are fixed with training done only on the state readout. Beside the advantages of ESN, there remains some opacity in the dynamic properties of the reservoir due to the presence of randomisation. Our aim in this paper is to demystify the model of ESN in a complete deterministic structure with the use of different proposed reservoir structures (topologies) and compare their performance with the random ESN on different benchmark datasets. All applied topologies maintain the simplicity of random ESN computation complexity. Most of the topologies showed comparable or even better performance.

Metal oxide gas sensors are important in industrial fields. Over the time, gas sensors tend to show inaccurate readings, therefore sensor drift arise causing a weakness in analyzing and identifying the correct type of gases. Different machine learning models implemented to correctly classifying and identifying the gases with the existence of drift problem. In this paper, a classification model for identifying the correct class for six different gases is implemented. Two algorithms are used for the purpose of classification: Decision trees and Neural Networks. Both algorithms were found to have high accuracy in classification when applied to the dataset under study.

Cloud Computing is an emerging technology which aims to provide efficient access to remote and distributed resources with the help of cloud services providers such as Amazon Googles Application, Microsoft Azure. Storage as a Service (STaaS) is among the most important and popular kind of services being investigated in the Cloud computing community to serve both private users and business enterprise with remote and scalable data storage that can be charged per customer usage. Providing safety, privacy and availability for the data that is hosted in cloud are the major concerns for users and businesses. This survey presents the data storage security concerns that arise due to the nature of cloud computing like multi-tenancy and virtualization, in addition to the details of the recent solutions presented in the literature to overcome the security issues in cloud storage.

Due to the growing use of social media, incidents of online abuse are also on rise. Online abusive behavior is defined as the use of electronic devices connected through internet for offensive activities. It is mostly in the form of comments containing abusive words about others, which affect the target users’ psychology and depresses them. This paper is aimed at devising method for detecting abusive behavior using supervised learning techniques. Two hypotheses are presented to extract features for detection of offensive comments. The initial experiments show that using features using our proposed method has better accuracy than the traditional feature extraction techniques like TF-IDF.

Churn prediction represents one of the most important components of Customer Relationship Management (CRM). In the purpose of retaining customers and maintaining their satisfaction, researchers of many fields including business intelligence, marketing and information technology were motivated to investigate the best methods that deliver the best services for customers. Many machine learning algorithms had been implemented in the purpose of optimally predicting the possible churning customers and making the right decisions at the right moments. Researchers had conducted several studies on various types of algorithms and results were found very promising. In this paper, we are conducting a comparison study of the performance towards churn prediction between two of the most powerful machine learning algorithms which are Decision Tree and K-Nearest Neighbor algorithms. Results were quite interesting showing a quite large dissimilarity in many areas between the two algorithms.