Experimental Evaluation Research Articles

Two algorithms for improving model-based diagnosis using multiple observations and deep learning.

Model-based diagnosis (MBD) is a critical problem in artificial intelligence. Recent advancements have made it possible to address this challenge using methods like deep learning. However, current approaches that use deep learning for MBD often struggle with accuracy and computation time due to the limited diagnostic information provided by a single observation. To address this challenge, we introduce two novel algorithms, Discret2DiMO (Discret2Di with Multiple Observations) and Discret2DiMO-DC (Discret2Di with Multiple Observations and Dictionary Cache), which enhance MBD by integrating multiple observations with deep learning techniques. Experimental evaluations on a simulated three-tank model demonstrate that Discret2DiMO significantly improves diagnostic accuracy, achieving up to a 685.06% increase and an average improvement of 59.18% over Discret2Di across all test cases. To address computational overhead, Discret2DiMO-DC additionally implements a caching mechanism that eliminates redundant computations during diagnosis. Remarkably, Discret2DiMO-DC achieves comparable accuracy while reducing computation time by an average of 95.74% compared to Discret2DiMO and 89.42% compared to Discret2Di, with computation times reduced by two orders of magnitude. These results indicate that our proposed algorithms significantly enhance diagnostic accuracy and efficiency in MBD compared with the state-of-the-art algorithm, highlighting the potential of integrating multiple observations with deep learning for more accurate and efficient diagnostics in complex systems.

Experimental and numerical evaluation of the green sand compaction process

ABSTRACT Determining the parameters of the sand moulding process plays an important role in obtaining green sand mould without wasting the energy used in the moulding process. The present work focuses on studying the variation effect of the sand density and elasto-plastic properties on the compaction process of sand moulding. Triaxial and uniaxial compression tests were used to estimate the properties of bentonite-bonded green sand. The behaviour of the green sand mould during the compaction process was investigated numerically by commercial finite element software, and experimentally using a flowability sensor. Detection of the green sand movement by a flowability sensor showed good agreement between the experimental and numerical results. Similar behaviour was observed for the distance-dependent displacement decrease with increasing green sand density. Stress concentration has an effect on the sand flow, density distribution and displacement of green mould sand.

Robot-Assisted CT-guided Biopsy with an Artificial Intelligence-Based Needle-Path Generator: An Experimental Evaluation Using a Phantom Model.

To investigate the feasibility of a robotic system with artificial intelligence-based lesion detection and path planning for CT-guided biopsy, compared to the conventional freehand technique. Eight nodules within an abdominal phantom, incorporating the simulated vertebrae and ribs, were designated as targets. A robotic system was used for lesion detection, trajectory generation, and needle-holder positioning. Four interventional radiologists with over 5 years of experience and four with 5 years or less performed 96 robot-assisted insertions encompassing both in-plane and out-of-plane trajectories. Additionally, 32 CT fluoroscopy-guided freehand needle insertions were performed along the in-plane trajectories. The three-dimensional (3D), lateral, depth deviations, and insertion time were quantified using post-needle insertion CT scans. Statistical analysis was performed using the unpaired t-test or one-way analysis of variance, with a significance level of P<0.05. The system detected all the target lesions and generated appropriate needle paths. Robot-assisted insertions exhibited significantly smaller 3D and depth deviations than freehand insertions (3.8 ± 1.3 vs. 4.7 ± 1.6 mm, P = 0.001 and 1.8 ± 1.2 vs. 2.6 ± 1.8 mm, P = 0.005, respectively). No significant difference was observed in lateral deviations (3.0 ± 1.5 vs. 3.5 ± 1.5 mm, P = 0.118). Robotic assistance significantly reduced insertion time compared to freehand insertion (17.3 ± 7.8 vs. 78.6 ± 38.1 sec, P < 0.001). The same trends were observed between the two groups of radiologists. The robotic system has the potential to shorten puncture time while maintaining sufficient accuracy in CT-guided procedures.

Experimental Evaluation of a Mobile Charging Station Prototype for Energy Supply Applied to Rural and Isolated Areas in Emergency Situations

Experimental Evaluation of a Mobile Charging Station Prototype for Energy Supply Applied to Rural and Isolated Areas in Emergency Situations

Experimental evaluation of residual tensile properties of CFRP‐aluminum butt adhesive joints subjected to systematic creep damage

Experimental evaluation of residual tensile properties of <scp>CFRP</scp>‐aluminum butt adhesive joints subjected to systematic creep damage

“Telephone Angels” Against Loneliness: Experimental Evaluation of the Effectiveness of Telephone Partnerships with Older Adults

ABSTRACT Loneliness affects many older adults.As part of the ”Telephone Angel” project, telephone partnerships between volunteers and older adults affected by loneliness were designed to counteract experiencing loneliness. Volunteers (100 ≤ N ≤ 114) and older adults who are (22 ≤ N ≤ 45) and who are not (25 ≤ N ≤ 71) part of the project were surveyed twice. Concerning loneliness, telephone partnerships increased the sense of community (d = .38). Older adults' life satisfaction increased (d = .46) as well. Stigmatization increased between the survey periods for those inside and outside the project (.21 ≤ d ≤ .35)..

Hierarchical Membrane Computing Algorithms for Optimizing Customer-to-Green-Manufacturer Decision-Making in Industrial Internet Platforms

This paper proposes a dynamic membrane algorithm (DMA)-oriented computing framework designed to optimize decision-making in Customer-to-Green-Manufacturer (C2GM) operations on industrial internet platforms. Unlike traditional methods that focus solely on economic metrics, the DMA integrates membrane computing principles with evolutionary optimization techniques and incorporates green manufacturing objectives (e.g., energy efficiency, waste reduction, carbon footprint). By doing so, it dynamically aligns customer demands with manufacturing capabilities in real time, ensuring both operational efficiency and environmental stewardship. The DMA facilitates parallel and hierarchical processing of complex decision tasks, mapping evolutionary rules and manufacturing operations into a structured membrane system that accelerates convergence and improves scalability. Experimental evaluations—both in controlled simulations and a real-world case study of C2GM operations in Yiwu—demonstrate that the DMA not only achieves faster and more stable convergence than genetic algorithms but also supports greener production processes. This integrated approach thus enhances strategic decision-making, offering a sustainable pathway for advancing industrial internet ecosystems and global supply chains.

Applying an Equity Lens to Evidence-Based Preventive Interventions: a Systematic Review of Subgroup Findings from Experimental Evaluations.

Evidence reveals that minoritized groups face disparities, underscoring the need for interventions to address behavioral health inequities. This review examined which minoritized populations are represented in evidence-based preventive interventions (EBPIs) and whether they equitably benefit from these programs. Using the Blueprints for Healthy Youth Development online clearinghouse, we synthesized findings from 240 high-quality experimental evaluations of EBPIs conducted in the U.S. between 2010 and 2023 and performed a descriptive analysis based on consensus coding to assess (1) the prevalence of culturally tailored EBPIs; (2) how frequently tests for subgroup effects were conducted; and (3) whether subgroup tests indicated differential benefits for minoritized groups. We found few culturally tailored interventions (31%), with 4% evaluating EBPIs developed for African American or Black populations and 1% for Hispanic or Latino youth. Additionally, only 25% and 15% tested for subgroup effects by race and ethnicity, respectively. For other subgroups, few (28%) evaluations included effects by economic disadvantage while 47% examined outcomes by binary gender categories. Essentially no reports tested for subgroup effects by sexual identity, location, or nativity status. Encouraging findings were that EBPIs more often benefited racial and ethnic minoritized groups, and there was an upward trend in reporting subgroup tests across time. EBPIs should test for subgroup effects to answer the questions of "what works for whom?" and "in which settings?" and to better understand the generalizability of findings. Investments are needed in culturally grounded programs developed for historically marginalized populations and trials of EBPIs that investigate mitigating health disparities.

Efficient and Secure Traffic Scheduling Based on Private Sketch

In today’s data–driven world, the explosive growth of network traffic often leads to network congestion, which seriously affects service performance and user experience. Network traffic scheduling is one of the key technologies to deal with congestion problems. Traditional traffic scheduling methods often rely on static rules or pre–defined policies, which make it difficult to cope with dynamically changing network traffic patterns. Additionally, the inability to efficiently manage tail contributors that disproportionately contribute to traffic can further exacerbate congestion issues. In this paper, we propose ESTS, an efficient and secure traffic scheduling based on private sketch, capable of identifying tail contributors to adjust routing and prevent congestion. The key idea is to develop a randomized admission (RA) structure, linking two count–mean–min (CMM) sketches. The first CMM sketch records cold items, while the second, following the RA structure, stores hot items with high frequency. Moreover, considering that tail contributors may leak private information, we incorporate Gaussian noise uniformly into the CMM sketch and RA structure. Experimental evaluations on real and synthetic datasets demonstrate that ESTS significantly improves the accuracy of feature distribution estimation and privacy preservation. Compared to baseline methods, the ESTS framework achieves a 25% reduction in average relative error and a 30% improvement in tail contributor identification accuracy. These results underline the framework’s efficiency and reliability.

Dynamic configuration optimization of FPGA accelerators through reinforcement learning for enhanced performance and resource utilization

Abstract Deploying Sparse Ternary Neural Networks on edge devices is in the areas of computational efficiency and energy optimization is a challenging task. This work presents a new FPGA-based accelerator integrating reinforcement learning and neural architecture search to dynamically optimize Sparse Ternary Neural Networks (Sparse TNN) for real-time applications. The design adopts adaptive pruning and quantization techniques for computational complexity and power consumption with the desired accuracy. Experimental evaluation on the Xilinx ZCU102 platform achieves up to 16.46× speedup compared to dense models at less than 1% accuracy loss and achieves state-of-the-art performance on benchmarks such as Google Net and MobileNetV2. This work holds promise for resource-constrained high-throughput applications, bringing FPGA-based deep learning closer to efficiency and scalability.

On the Practicability of Deep Learning based Anomaly Detection for Modern Online Software Systems: A Pre-Train-and-Align Framework

Operation and maintenance are critical activities in the whole life cycle of modern online software systems, and anomaly detection is a crucial step of these activities. Recent studies mainly develop deep learning techniques to complete this task. Notably, though these techniques have achieved promising results in experimental evaluations, there are still several practicality gaps for them to be successfully applied in a real-world online system, including the scalability gap, availability gap and alignment gap. To bridge these gaps, we propose an anomaly detection framework, namely ShareAD , based on a pre-train-and-align paradigm. Specifically, we argue that pre-training a shared model for anomaly detection is an effective way to bridge the scalability gap and the availability gap. To support this argument, we systematically study the necessity and feasibility of model sharing for online system maintenance. We further propose a novel model based upon Transformer encoder layers and Base layers, which works well for anomaly detection pre-training. Then, to bridge the alignment gap, we propose ShareAD alignment to align the pre-trained model with operator preference by jointly considering the local observation context and sensitivity of each monitor entity. Extensive experiments on two real-world large-scale datasets demonstrate the effectiveness and practicality of ShareAD .

Research on Advertising Prediction Based on Stacking Ensemble Algorithms

The purpose of advertising prediction is to analyze user behavior and features to forecast the effectiveness of ad placements, thereby achieving a higher return on investment. This paper employs a stacking-based ensemble learning method, integrating multiple machine learning algorithms to enhance predictive capabilities. To support this method, this article selected a dataset related to advertising click through rate (CTR). The paper constructs two models: Gradient Boosting Decision Tree (GBDT) and eXtreme Gradient Boosting (XGBoost). Then the paper utilizes Bayesian optimization to tune the model parameters to achieve optimal parameter combinations. The paper uses these two individual models as base learners in a stacking ensemble model, with a random forest model serving as the meta-learner for training. This paper evaluates the predictive performance of the models using two metrics: log loss value, and AUC value. After experimental evaluation, the AUC value of the Stack fusion model reached 0.9999, which is about 0.42%, 1.29%, and 10.99% higher than XGBoost (0.9956), Random Forest (0.9871), and GBDT (0.8904), respectively. In terms of logloss value, the Stack fusion model has a loss of 0.0101, significantly lower than XGBoost's 0.0595, a decrease of about 83.01%. Compared to the random forest's 0.1896 and GBDT's 0.1884, it decreased by approximately 94.67% and 94.64%, respectively. This result indicates that the ensemble model proposed in this paper demonstrates a significant improvement in click-through rate prediction when contrasted with individual models.

Study of Baclofen Solubility in Supercritical CO2 with and without Cosolvents: Experimental Analysis, Thermodynamic Evaluation, and Machine Learning Methods

Study of Baclofen Solubility in Supercritical CO₂ with and without Cosolvents: Experimental Analysis, Thermodynamic Evaluation, and Machine Learning Methods

Experimental Evaluation of the Response of AlGaN/GaN UV Photodetectors With Square and Hexagonal Nanohole Arrays

Experimental Evaluation of the Response of AlGaN/GaN UV Photodetectors With Square and Hexagonal Nanohole Arrays

A detailed experimental and theoretical evaluations of two thiazolidine-2, 4-dione derivatives as corrosion inhibitors for carbon steel in a hydrochloric acid electrolyte

In this study, we conducted a comparative analysis of the inhibiting effect of two recently developed thiazolidine-2,4-dione derivatives, namely (Z)-3-allyl-5-(4-bromobenzylidene) thiazolidine-2,4-dione (TZD1) and (Z)-5-(4-bromobenzylidene) thiazolidine-2,4-dione (TZD2), on the corrosion protection of carbon steel (Csteel) in a 1 M HCl electrolyte. Both potentiodynamic polarization (PDP) and electrochemical impedance spectroscopy (EIS) techniques were used to do this comparison. The results indicate that the concentrations of TZD1 and TZD2, as well as the experimental temperature, are important factors that affect inhibitory efficacy. At 303 K and a concentration of 10−3M, the highest significant inhibition rates were achieved, reaching 93.7% for TZD1 and 91.7% for TZD2. The PDP analysis revealed that TZD1 and TZD2 are mixed-type inhibitors that primarily function as anodic inhibitors. Using the Langmuir isotherm model as a guide, TZD1 and TZD2 molecules chemisorb onto the steel surface. Surface characterization analysis utilizing scanning electron microscopy (SEM), contact angle (CA), and X-ray diffraction (XRD) confirmed this interaction. Additionally, UV-visible spectra showed that TZD1 and TZD2 molecules interact significantly with Fe ions at certain atomic positions. To investigate how the molecular structure of TZD1 and TZD2 affects their ability to inhibit corrosion, Fukui functions, Molecular Dynamics Simulations (MDS), and DFT calculations were employed. An examination was also conducted to determine how the thiazolidine derivatives protonate in an acidic environment. Strong agreement was observed between the outcomes of these different approaches.

Microscopic augmented reality calibration with contactless line-structured light registration for surgical navigation.

The use of AR technology in image-guided neurosurgery enables visualization of lesions that are concealed deep within the brain. Accurate AR registration is required to precisely match virtual lesions with anatomical structures displayed under a microscope. The purpose of this work was to develop a real-time augmented surgical navigation system using contactless line-structured light registration, microscope calibration, and visible optical tracking. Contactless discrete sparse line-structured light point cloud is utilized to construct patient-image registration. Microscope calibration optimization with dimensional invariant calibrator is employed to enable real-time tracking of the microscope. The visible optical tracking integrates a 3D medical model with surgical microscope video in real time, generating an augmented microscope stream. The proposed patient-image registration algorithm yielded an average root mean square error (RMSE) of 0.78 ± 0.14mm. The pixel match ratio error (PMRE) of the microscope calibration was found to be 0.646%. The RMSE and PMRE of the system experiments are 0.79 ± 0.10mm and 3.30 ± 1.08%, respectively. Experimental evaluations confirmed the feasibility and efficiency of microscope AR surgical navigation (MASN) registration. By means of registration technology, MASN overlays virtual lesions onto the microscopic view of the real lesions in real time, which can help surgeons to localize lesions hidden deep in tissue.

Experimental evaluation of DC-DC buck converter based on adaptive fuzzy fast terminal synergetic controller

This study suggests an enhanced version of the adaptive fuzzy fast terminal synergetic controller (AF-FTSC) for controlling the uncertain DC/DC buck converter based on the synergetic theory of control (STC) and newly developed terminal attractor technique (TAT). The benefits of the proposed SC algorithm involve the features of finite-time convergence, unaffected by parameter variations, and chattering-free phenomenon. A type-1 fuzzy logic system (T1-FLS) make the considered controller more robust and is utilized to estimate the undefined converter nonlinear dynamics without resorting to the usual linearization and simplifications of the converter model. Taking a switching DC-DC buck converter as a demonstration, the suggested AF-FTSC is thoroughly analyzed and executed on a dSPACE ds1103 controller board. The outcomes of the experiment confirm the competence and applicability of the suggested regulator.

Mechanically Robust Bismuth-Embedded Carbon Microspheres for Ultrafast Charging and Ultrastable Sodium-Ion Batteries.

Advancements in the development of fast-charging and long-lasting microstructured alloying anodes with high volumetric capacities are essential for enhancing the operational efficiency of sodium-ion batteries (SIBs). These anodes, however, face challenges such as declined cyclability and rate capability, primarily due to mechanical degradation reduced by significant volumetric changes (over 252%) and slow kinetics of sodium-ion storage. Herein, we introduce a novel anode design featuring densely packed bismuth (Bi) embedded within highly conductive carbon microspheres to overcome the aforementioned challenges. Remarkably, the high loading Bi anode within carbon microspheres with a high tap density of 2.59 g cm-3 possesses significant mechanical strength exceeding 590 MPa and limits volume swelling of only 10.9% post-sodiation. This anode demonstrates a high volumetric capacity (908.3 mAh cm-3), ultrafast chargeability (200 A g-1, full charge/discharge in just 5.5 s), and outstanding cyclability over 12,000 cycles and maintains exceptional cycling stability even at -30 °C. The full cell paired with a Na3V2(PO4)3 cathode retains over 80% capacity after 600 cycles at 36 C, demonstrating a remarkable rate capability of 126 C (full charge/discharge in 28.6 s). Our comprehensive experimental evaluations and chemo-mechanical simulations shed light on the mechanisms underpinning the anode's superior performance. This development marks a significant advancement in the design of durable and fast-charging anodes for high-performance SIBs.

Surface State Control of Apatite Nanoparticles by pH Adjusters for Highly Biocompatible Coatings.

Apatite nanoparticles are biocompatible nanomaterials, so their film formation on biodevices is expected to provide effective bonding with living organisms. However, the biodevice-apatite interfaces have not yet been elucidated because there is little experimental evaluation and discussion on the nanoscale interactions, as well as the apatite surface reactivities. Our group has demonstrated the biomolecular adsorption properties on a quartz crystal microbalance with dissipation (QCM-D) sensor coated with apatite nanoparticles, demonstrating the applicability of apatite nanoparticle films on devices. Therefore, it is important to clarify the biodevice-apatite nanointerfaces by characterizing their physicochemical properties. This research aims to control the apatite nanoparticle surfaces using different types of pH adjusters as well as to investigate biodevice-apatite interfaces. In this study, tetramethylammonium hydroxide, sodium hydroxide, and potassium hydroxide were used to adjust the pH during the synthesis of apatite nanoparticles. As a result, it was found that the ratio of Ca-deficient hydroxyapatite phase to B-type carbonate ion-substituted hydroxyapatite phase could be controlled by adjusting the OH- concentration and that the formation of B-type carbonate ion substituted hydroxyapatite phase was demonstrated in terms of the charge compensation because hydrogen phosphate ion in the non-apatitic layer would be diffused and substituted inside the apatite core structure by the replacement of carbonate ion. By contrast, the phosphate ions in the core structure were moved and contained in the non-apatitic layer, and the proportion of phosphate ions in the non-apatitic layer increased. The surface changes of the nanoparticles provide an effective biodevice surface coating. It was observed that the thickness of the electrophoretically deposited nanoparticles clearly increased with the proportion of phosphate ions in the non-apatitic layer. Furthermore, the formation of the hydration layer with immersion in biological fluid was evaluated. It was inferred that the water molecules in the hydration layer interacted with the substituted ions and remained as nonfreezing water layer on the top surface of the nanoparticles, while the abundant phosphate ions newly interacted with the water molecules in the non-apatitic layer, thus increasing the proportion of intermediate water. These results indicated that the hydrogen phosphate and phosphate ions were retained in the non-apatitic layer on the top surfaces of apatite nanoparticles, so that the thickness of the electrophoretically deposited film and the weight fraction of the hydrated layer can be controlled by the component ratio of phosphate ions in the non-apatitic layer. It is expected that surface coating technology using apatite nanoparticles will be applied for biodevices.

Research on the construction of intelligent-aided design platform for green buildings

While socio-economic development and technological advancement bring great convenience to our life, energy consumption and environmental pollution have also become the core issues affecting human survival and development. This study focuses on the development of an intelligent design platform for green buildings, aimed at promoting sustainability through low energy consumption and reduced environmental impact. The study of existing national standards and requirements for the intelligent design of green buildings, combined with the unique characteristics of green buildings and user needs, has led to the development of an optimized auxiliary design platform system. By incorporating algorithm optimization and modeling techniques, this platform enhances the design process. Experimental evaluations of the system reveal that the intelligent auxiliary design offers more accurate predictions, thereby improving user service and experience. Additionally, the system contributes to energy efficiency and environmental sustainability by reducing the building’s ecological footprint. This research highlights the potential of intelligent systems to facilitate more efficient and environmentally friendly green building designs.