The Partial Max-SAT (PMSAT) problem is an optimization variant of the well-known Propositional Boolean Satisfiability (SAT) problem. It holds an important place in theory and practice, because a huge number of real-world problems, such as timetabling, planning, routing, bioinformatics, fault diagnosis, etc., could be encoded into it. Stochastic local search (SLS) methods can solve many real-world problems that often involve large-scale instances at reasonable computation costs while delivering good-quality solutions. In this work, we propose a novel SLS algorithm called adaptive variable depth SLS for PMSAT problem solving based on a dynamic local search framework. Our algorithm exploits two algorithmic components of an SLS method: parameter tuning and neighborhood search. Our first contribution is the design of an adaptive parameter tuner that searches for the best parameter setting for each instance by considering its features. The second contribution is a variable depth neighborhood search (VDS) algorithm adopted for PMSAT problem, which our empirical evaluation proves is a more efficient w.r.t. single neighborhood search. We conducted our experiments on the PMSAT benchmarks from MaxSAT Evaluation 2014 to 2019, including more than 3600 instances which have been encoded from a broad range of domains such as verification, optimization, graph theory, automated-reasoning, pseudo Boolean, etc. Our experimental evaluation results show that AVD-SLS solver, which is implemented based on our algorithm, outperforms state-of-the-art PMSAT SLS solvers in most benchmark classes, including random, crafted, and industrial instances. Furthermore, AVD-SLS reports remarkably better results on weighted benchmark, and shows competitive results with several well-known hybrid PMSAT solvers.

This study presents a comprehensive evaluation of machine learning (ML) methods for the early detection of depression in Arabic tweets, addressing critical gaps in mental health informatics for Arabic-speaking populations. We introduce the ArabMindGuard (AMG) dataset, which comprises 3083 expertly annotated tweets from Saudi Arabia, and perform systematic comparisons across five datasets that encompass multiple Arabic dialects, including Saudi, Jordanian, and Modern Standard Arabic (MSA). Our research employs a rigorous evaluation framework, incorporating multiple performance metrics such as precision, recall, F1-score, and statistical significance testing. Leveraging advanced Arabic-specific natural language processing (NLP) techniques–including Term Frequency-Inverse Document Frequency (TF-IDF), N-gram analysis, and CAMeL tools–we assess the performance of nine machine learning classifiers with statistical rigor. The results show that ensemble methods and Support Vector Machines (SVM) consistently outperform other approaches, achieving F1-scores ranging from 0.82 to 0.95 across the datasets. These findings highlight the significant role of dialectal variation in depression detection and provide foundational baseline metrics for future deep learning applications. Overall, this study makes a valuable contribution to culturally responsive digital mental health interventions in the Arab world and establishes methodological benchmarks for Arabic mental health text analysis.

Obstructive Sleep Apnea (OSA) is a prevalent respiratory disorder traditionally diagnosed with costly and inaccessible overnight polysomnography. While single-lead electrocardiogram (ECG) signals offer a scalable alternative for automated screening, existing deep learning models often struggle with multi-class severity grading and cross-dataset generalization. In this study, we propose a robust, ECG-based hybrid deep learning framework built upon a baseline CNN-LSTM architecture. Through a rigorous ablation study on the PhysioNet Apnea-ECG dataset testing nine distinct optimizations, we demonstrate that combining the sequential CNN-LSTM with a multi-scale inception-style CNN coupled with physiologically motivated data augmentation and Youden Index threshold calibration improves patient-level severity classification accuracy from 78.6% to 89.3%. The final ensemble achieved a segment-level accuracy of 91.76%, a specificity of 94.6%, and an Area Under the Curve (AUC) of 0.965. Furthermore, we conducted a systematic cross-dataset evaluation on the MIT-BIH Polysomnographic Database to isolate the effects of domain shift and segmentation window variations. Comprehensive feature-space analysis revealed that while hybrid architectures including Random Forest classifiers improve latent feature discrimination, they remain highly vulnerable to sensitivity collapse due to threshold miscalibration on compressed target-domain probability distributions. These findings underscore the necessity of target-domain threshold recalibration for the clinical deployment of ECG-based OSA diagnostic support tools.