Search Results (244)

In the real world, most data are unlabeled, which drives the development of semi-supervised learning (SSL). Among SSL methods, least squares regression (LSR) has attracted attention for its simplicity and efficiency. However, existing semi-supervised LSR approaches suffer from challenges such as the insufficient use of unlabeled data, low pseudo-label accuracy, and inefficient label propagation. To address these issues, this paper proposes dual label propagation-driven least squares regression with feature selection, named DLPLSR, which is a pseudo-label-free SSL framework. DLPLSR employs a fuzzy-graph-based clustering strategy to capture global relationships among all samples, and manifold regularization preserves local geometric consistency, so that it implements the dual label propagation mechanism for comprehensive utilization of unlabeled data. Meanwhile, a dual-feature selection mechanism is established by integrating orthogonal projection for maximizing feature information with an ℓ_2,1-norm regularization for eliminating redundancy, thereby jointly enhancing the discriminative power. Benefiting from these two designs, DLPLSR boosts learning performance without pseudo-labeling. Finally, the objective function admits an efficient closed-form solution solvable via an alternating optimization strategy. Extensive experiments on multiple benchmark datasets show the superiority of DLPLSR compared to state-of-the-art LSR-based SSL methods. Full article

Image dehazing technology is a crucial component in the fields of intelligent transportation and autonomous driving. However, most existing dehazing algorithms only process images in the spatial domain, failing to fully exploit the rich information in the frequency domain, which leads to residual haze in the images. To address this issue, we propose a novel Frequency-domain Information Guided Symmetric Dual-branch Dehazing Network (FIGD-Net), which utilizes the spatial branch to extract local haze features and the frequency branch to capture the global haze distribution, thereby guiding the feature learning process in the spatial branch. The FIGD-Net mainly consists of three key modules: the Frequency Detail Extraction Module (FDEM), the Dual-Domain Multi-scale Feature Extraction Module (DMFEM), and the Dual-Domain Guidance Module (DGM). First, the FDEM employs the Discrete Cosine Transform (DCT) to convert the spatial domain into the frequency domain. It then selectively extracts high-frequency and low-frequency features based on predefined proportions. The high-frequency features, which contain haze-related information, are correlated with the overall characteristics of the low-frequency features to enhance the representation of haze attributes. Next, the DMFEM utilizes stacked residual blocks and gradient feature flows to capture local detail features. Specifically, frequency-guided weights are applied to adjust the focus of feature channels, thereby improving the module’s ability to capture multi-scale features and distinguish haze features. Finally, the DGM adjusts channel weights guided by frequency information. This smooths out redundant signals and enables cross-branch information exchange, which helps to restore the original image colors. Extensive experiments demonstrate that the proposed FIGD-Net achieves superior dehazing performance on multiple synthetic and real-world datasets. Full article

This work systematically addresses the dual challenges of non-inertial dynamic coupling and kinematic constraint redundancy encountered in dynamic modeling of serial–parallel–serial hybrid robotic mechanisms, and proposes an improved Newton–Euler modeling method with constraint compensation. Taking the Skiing Simulation Platform with 6-DOF as the research mechanism, the inverse kinematic model of the closed-chain mechanism is established through $G_F$ set theory, with explicit analytical expressions derived for the motion parameters of limb mass centers. Introducing a principal inertial coordinate system into the dynamics equations, a recursive algorithm incorporating force/moment coupling terms is developed. Numerical simulations reveal a 9.25% periodic deviation in joint moments using conventional methods. Through analysis of the mechanism’s intrinsic properties, it is identified that the lack of angular momentum conservation constraints on the end-effector in non-inertial frames leads to system controllability degradation. Accordingly, a constraint compensation strategy is proposed: establishing linearly independent differential algebraic equations supplemented with momentum/angular momentum balance equations for the end platform. Co-Simulation results demonstrate that the optimized model reduces the maximum relative error of actuator joint moments to 0.98%, and maintains numerical stability across the entire configuration space. The constraint compensation framework provides a universal solution for dynamics modeling of complex closed-chain mechanisms, validated through applications in flight simulators and automotive driving simulators. Full article

Dual-motor electric vehicles enhance power performance and overall output capabilities by enabling the real-time control of the torque distribution between the front and rear wheels, thereby improving handling, stability, and safety. In addition to increased energy efficiency, a dual-motor system provides redundancy: if one motor fails, the other can still supply partial power, further enhancing driving safety. This study aimed to optimize the energy management strategies of the front- and rear-axis motors, examining the application effects of rule-based control (RBC), global grid search (GGS), and the whale optimization algorithm (WOA). A simulation platform based on MATLAB/Simulink^® (R2021b, MATLAB, Natick, MA, USA) was constructed and validated through hardware-in-the-loop (HIL) testing to ensure the authenticity and reliability of the simulation results. Detailed tests and analyses of the dual-motor system were conducted under FTP-75 driving cycles. Compared to the RBC strategy, GGS and WOA achieved energy efficiency improvements of 9.1% and 8.9%, respectively, in the pure simulation, and 4.2% and 3.8%, respectively, in the HIL simulation. Compared to the pure RBC strategy, the RBC and GGS strategies incorporating regenerative braking achieved energy efficiency improvements of 26.1% and 29.4%, respectively, in the HIL simulation. Overall, GGS and WOA each present distinct advantages, with WOA emerging as a highly promising alternative energy management strategy. Future research should further explore WOA applications to enhance energy savings in real-world vehicle operations. Full article

Precise grading of manufactured sand is vital to concrete performance, yet standard sieve tests, though accurate, are too slow for online quality control. Thus, we devised an image-based inspection method combining a dual-camera module with a Temporal Interval Sampling Strategy (TISS) to enhance throughput while maintaining precision. In this design, a global wide-angle camera captures the entire particle field, whereas a local high-magnification camera focuses on fine fractions. TISS selects only statistically representative frames, effectively eliminating redundant data. A lightweight segmentation algorithm based on geometric rules cleanly separates overlapping particles and assigns size classes using a normal-distribution classifier. In tests on ten 500 g batches of manufactured sand spanning fine, medium, and coarse gradations, the system processed each batch in an average of 7.8 min using only 34 image groups. It kept the total gradation error within 12% and the fineness-modulus deviation within ±0.06 compared to reference sieving. These results demonstrate that the combination of complementary optics and targeted sampling can provide a scalable, real-time solution. Full article

In the context of the rapid digital transformation of the catering industry, the design of ordering interfaces—key hubs of human–computer interaction—has become critical to user service quality and brand competitiveness, especially in terms of usability, visual appeal, and emotional resonance. Based on a human–computer interaction design framework, this study proposes a dual-path optimization model integrating user demand mapping and conflict resolution to synergize explicit need translation with innovative design problem solving. The model employs KE to capture implicit user needs, applies AHP to construct a weighted design element system, and uses QFD to establish a matrix linking user needs with technical attributes. To address contradictions among design elements, TRIZ is introduced to resolve conflicts between functional redundancy and interaction efficiency. Additionally, generative AI tools such as MidJourney are incorporated to accelerate concept generation and improve innovation. Based on user evaluations and controlled experiments, the optimized design demonstrates measurable improvements in task efficiency and visual appeal. Overall, the dual-path approach effectively bridges the gap between vague user needs and concrete design solutions, achieving a balanced integration of functionality, aesthetics, and interactivity. The proposed model overcomes the limitations of experience-driven design by offering a systematic methodology encompassing demand analysis, technological transformation, conflict resolution, and intelligent generation, with practical value for enhancing the user experience of digital service touchpoints in the catering sector. Full article

Recent advancements in decoding electroencephalogram (EEG) signals for motor imagery tasks have shown significant potential. However, the intricate time–frequency dynamics and inter-channel redundancy of EEG signals remain key challenges, often limiting the effectiveness of single-scale feature extraction methods. To address this issue, we propose the Dual-Branch Blocked-Integration Self-Attention Network (DB-BISAN), a novel deep learning framework for EEG motor imagery classification. The proposed method includes a Dual-Branch Feature Extraction Module designed to capture both temporal features and spatial patterns across different scales. Additionally, a novel Blocked-Integration Self-Attention Mechanism is employed to selectively highlight important features while minimizing the impact of redundant information. The experimental results show that DB-BISAN achieves state-of-the-art performance. Also, ablation studies confirm that the Dual-Branch Feature Extraction and Blocked-Integration Self-Attention Mechanism are critical to the model’s performance. Our approach offers an effective solution for motor imagery decoding, with significant potential for the development of efficient and accurate brain–computer interfaces. Full article

Several of the integrin family of cell adhesion receptors have been popular targets for the development of anticancer agents, but with little clinical success to date. Cancer cells usually express multiple redundant integrins; one hypothesis for the lack of efficacy of current antagonists is their high selectivity for a single integrin. To address this, we developed a functional dual-β3-expressing cell model to investigate the effects of combined αIIbβ3/αvβ3 antagonism. We established that treating K562 chronic myeloid leukemia cells with 0.04 μM phorbol 12-myristate 13-acetate (PMA) for 40 h significantly upregulates functional αIIbβ3 and αvβ3 integrins. This optimized method provides a reliable platform for adhesion and detachment assays, enabling the characterization of dual integrin targeting strategies. Using this model, we demonstrate that combining αIIbβ3 and αvβ3 antagonists (GR144053 and cRGDfV) synergistically enhances inhibition of cell adhesion and promotes cell detachment compared to single-agent treatments. Our findings establish a reproducible approach for studying dual β3 integrin targeting, which can be used to investigate potential strategies for overcoming integrin redundancy in cancer therapeutics. Full article

Transcription factors (TFs) are proteins that control gene expression by binding to specific DNA sequences and are essential for cell development, differentiation, and homeostasis. Dysregulation of TFs is implicated in numerous diseases, including cancer, autoimmune disorders, and neurodegeneration. While TFs were traditionally considered “undruggable” due to their lack of well-defined binding pockets, recent advances have made it possible to modulate their activity using diverse pharmacological strategies. Major TF families include NF-κB, p53, STATs, HIF-1α, AP-1, Nrf2, and nuclear hormone receptors, which take part in the regulation of inflammation, tumor suppression, cytokine signaling, hypoxia and stress response, oxidative stress, and hormonal response, respectively. TFs can perform multiple functions, participating in the regulation of opposing processes depending on the context. NF-κB, for instance, plays dual roles in immunity and cancer, and is targeted by proteasome and IKKβ inhibitors. p53, often mutated in cancer, is reactivated using MDM2 antagonist Nutlin-3, refunctionalizing compound APR-246, or stapled peptides. HIF-1α, which regulates hypoxic responses and angiogenesis, is inhibited by agents like acriflavine or stabilized in anemia therapies by HIF-PHD inhibitor roxadustat. STATs, especially STAT3 and STAT5, are oncogenic and targeted via JAK inhibitors or novel PROTAC degraders, for instance SD-36. AP-1, implicated in cancer and arthritis, can be inhibited by T-5224 or kinase inhibitors JNK and p38 MAPK. Nrf2, a key antioxidant regulator, can be activated by agents like DMF or inhibited in chemoresistant tumors. Pharmacological strategies include direct inhibitors, activators, PROTACs, molecular glues, and epigenetic modulators. Challenges remain, including the structural inaccessibility of TFs, functional redundancy, off-target effects, and delivery barriers. Despite these challenges, transcription factor modulation is emerging as a viable and promising therapeutic approach, with ongoing research focusing on specificity, safety, and efficient delivery methods to realize its full clinical potential. Full article

Incorporating external knowledge has been shown to improve emotion understanding in dialogues by enriching contextual information, such as character motivations, psychological states, and causal relations between events. Filtering and categorizing this information can significantly enhance model performance. In this paper, we present an innovative Emotion Recognition in Conversation (ERC) framework, called the Scene-Speaker Emotion Awareness Network (SSEAN), which employs a dual-strategy modeling approach. SSEAN uniquely incorporates external commonsense knowledge describing speaker states into multimodal inputs. Using parallel recurrent networks to separately capture scene-level and speaker-level emotions, the model effectively reduces the accumulation of redundant information within the speaker’s emotional space. Additionally, we introduce an attention-based dynamic screening module to enhance the quality of integrated external commonsense knowledge through three levels: (1) speaker-listener-aware input structuring, (2) role-based segmentation, and (3) context-guided attention refinement. Experiments show that SSEAN outperforms existing state-of-the-art models on two well-adopted benchmark datasets in both single-text modality and multimodal settings. Full article

Efficient individual identification is essential for advancing precision broiler farming. In this study, we propose YOLO-IFSC, a high-precision and lightweight face recognition framework specifically designed for dense broiler farming environments. Building on the YOLOv11n architecture, the proposed model integrates four key modules to overcome the limitations of traditional methods and recent CNN-based approaches. The Inception-F module employs a dynamic multi-branch design to enhance multi-scale feature extraction, while the C2f-Faster module leverages partial convolution to reduce computational redundancy and parameter count. Furthermore, the SPPELANF module reinforces cross-layer spatial feature aggregation to alleviate the adverse effects of occlusion, and the CBAM module introduces a dual-domain attention mechanism to emphasize critical facial regions. Experimental evaluations on a self-constructed dataset demonstrate that YOLO-IFSC achieves a mAP@0.5 of 91.5%, alongside a 40.8% reduction in parameters and a 24.2% reduction in FLOPs compared to the baseline, with a consistent real-time inference speed of 36.6 FPS. The proposed framework offers a cost-effective, non-contact alternative for broiler face recognition, significantly advancing individual tracking and welfare monitoring in precision farming. Full article

Deep learning methods have achieved remarkable success in remote sensing object detection. Existing object detection methods focus on integrating convolutional neural networks (CNNs) and Transformer networks to explore local and global representations to improve performance. However, existing methods relying on fixed convolutional kernels and dense global attention mechanisms suffer from computational redundancy and insufficient discriminative feature extraction, particularly for small and rotation-sensitive targets. To address these limitations, we propose a Dynamic Multi-Kernel Position-Aware Feature Pyramid Network (PAMFPN), which integrates adaptive sparse position modeling and multi-kernel dynamic fusion to achieve robust feature representation. Firstly, we design a position-interactive context module (PICM) that incorporates distance-aware sparse attention and dynamic positional encoding. It selectively focuses computation on sparse targets through a decay function that suppresses background noise while enhancing spatial correlations of critical regions. Secondly, we design a dual-kernel adaptive fusion (DKAF) architecture by combining region-sensitive attention (RSA) and reconfigurable context aggregation (RCA). RSA employs orthogonal large-kernel convolutions to capture anisotropic spatial features for arbitrarily oriented targets, while RCA dynamically adjusts the kernel scales based on content complexity, effectively addressing scale variations and intraclass diversity. Extensive experiments on three benchmark datasets (DOTA-v1.0, SSDD, HWPUVHR-10) demonstrate the effectiveness and versatility of the proposed PAMFPN. This work bridges the gap between efficient computation and robust feature fusion in remote sensing detection, offering a universal solution for real-world applications. Full article

Microplastics and invasive species, driven by anthropogenic activities, significantly disrupt ecosystems and microbial communities. This study investigated the interactive effects of biodegradable microplastics (polylactic acid, or PLA, and polyhydroxyalkanoates, or PHAs) and the fungal pathogen Rhizoctonia solani on the invasive plant Solidago canadensis. One plant of Solidago canadensis/pot was cultivated in forest soil amended with 1% (w/w) microplastics and/or R. solani. PLA exhibited greater toxicity than PHAs, reducing the plant height, root length, and biomass by 68%, 44%, and 70%, respectively. Microplastics impaired the maximum quantum yield of photosystem II more severely than R. solani. However, S. canadensis demonstrated adaptive antioxidative and extracellular enzymatic mechanisms under combined stresses. A heatmap analysis revealed a positive correlation between PHAs and plant growth traits, while a redundancy analysis explained the 15.96% and 4.19% variability for the first two components (r² = 0.95). A structural equation model indicated the negative effects of morphology and physiology on biomass (β = −1.694 and β = −0.932; p < 0.001), countered by positive antioxidant contributions (β = 1.296; p < 0.001). These findings highlight complex interactions among microplastics, pathogens, and invasive species, offering insights into ecological management strategies under dual environmental pressures. Future studies should assess the long-term field effects and microbial mediation of these interactions. Full article

Hub motor-driven vehicles (HMDVs) suffer from poor handling and stability due to an increased unsprung mass and unbalanced radial electromagnetic forces. Although traditional ground-hook control reduces the dynamic tire load, it severely worsens the body acceleration. This paper presents a generalized ground-hook control strategy based on impedance transfer functions to address the parameter redundancy in structural methods. A quarter-vehicle model with a switched reluctance motor wheel hub drive was used to study different orders of generalized ground-hook impedance transfer function control strategies for dynamic inertial suspension. An enhanced fish swarm parameter optimization method identified the optimal solutions for different structural orders. Analyses showed that the third-order control strategy optimized the body acceleration by 2%, reduced the dynamic tire load by 8%, and decreased the suspension working space by 22%. This strategy also substantially lowered the power spectral density for the body acceleration and dynamic tire load in the low-frequency band of 1.2 Hz. Additionally, it balanced computational complexity and performance, having slightly higher complexity than lower-order methods but much less than higher-order structures, meeting real-time constraints. To address time-domain deviations from generalized ground-hook control in semi-active systems, a dynamic compensation strategy was proposed: eight topological structures were created by modifying the spring–damper structure. A deviation correction mechanism was devised based on the frequency-domain coupling characteristics between the wheel speed and suspension relative velocity. For ride comfort and road-friendliness, a dual-frequency control criterion was introduced: in the low-frequency range, energy transfer suppression and phase synchronization locking were realized by constraining the ground-hook damping coefficient or inertance coefficient, while in the high-frequency range, the inertia-dominant characteristic was enhanced, and dynamic phase adaptation was permitted to mitigate road excitations. The results show that only the T0 and T5 structures met dynamic constraints across the frequency spectrum. Time-domain simulations showed that the deviation between the T5 structure and the third-order generalized ground-hook impedance model was relatively small, outperforming traditional and T0 structures, validating the model’s superior adaptability in high-order semi-active suspension. Full article

As urban traffic systems become increasingly complex, traditional traffic data collection methods based on fixed detectors face challenges such as poor data quality and acquisition difficulties. Traditional methods also lack the ability to capture complete vehicle path information essential for signal optimization. While vehicle trajectory data can provide rich spatiotemporal information, its sampling characteristics present new technical challenges for traffic parameter extraction. This study addresses the key issue of extracting traffic parameters suitable for signal timing optimization from sampled trajectory data by proposing a comprehensive method for traffic parameter extraction and analysis based on vehicle trajectory data. The method comprises five modules: data preprocessing, basic feature processing, exploratory data analysis, key feature extraction, and data visualization. An innovative algorithm is proposed to identify which intersections vehicles pass through, effectively solving the challenge of mapping GPS points to road network nodes. A dual calculation method based on instantaneous speed and time difference is adopted, improving parameter estimation accuracy through multi-source data fusion. A highly automated processing toolchain based on Python and MATLAB is developed. The method advances the state of the art through a novel polygon-based trajectory mapping algorithm and a systematic multi-source parameter extraction framework specifically designed for signal control optimization. Validation using actual trajectory data containing 2.48 million records successfully eliminated 30.80% redundant data and accurately identified complete paths for 7252 vehicles. The extracted multi-dimensional parameters, including link flow, average speed, travel time, and OD matrices, accurately reflect network operational status, identifying congestion hotspots, tidal traffic characteristics, and unstable road segments. The research outcomes provide a feasible technical solution for areas lacking traditional detection equipment. The extracted parameters can directly support signal optimization applications such as traffic signal coordination, timing optimization, and congestion management, providing crucial support for implementing data-driven intelligent traffic control. This research presents a theoretical framework validated with real-world data, providing a foundation for future implementation in operational signal control systems. Full article