Compressible Model Research Articles

A Discrete-Mapping-Based Cross-Component Prediction Paradigm for Screen Content Coding.

Cross-component prediction is an important intra-prediction tool in the modern video coders. Existing prediction methods to exploit cross-component correlation include cross-component linear model and its extension of multi-model linear model. These models are designed for camera captured content. For screen content coding, where videos exhibit different signal characteristics, a cross-component prediction model tailored to their characteristics is desirable. As a pioneering work, we propose a discrete-mapping based cross-component prediction model for screen content coding. Our model relies on the core observation that, screen content videos typically comprise of regions with a few distinct colors and luma value (almost always) uniquely conveys chroma value. Based on this, the proposed method learns a discrete-mapping function from available reconstructed luma-chroma pairs and uses this function to derive chroma prediction from the co-located luma samples. To achieve higher accuracy, a multi-filter approach is employed to derive co-located luma values. The proposed method achieves 2.61%, 3.51% and 3.92% Y, U and V bit-rate savings respectively over Enhanced Compression Model (ECM) 4.0, with negligible complexity, for text and graphics media under all-intra configuration.

Cross-Component Prediction Boosted With Local and Non-Local Information in Video Coding.

In the versatile video coding (VVC) standard, cross-component prediction (CCP) is introduced to utilize the correlation between luma and chroma components. To further exploit the cross-component redundancy, a bundle of novel CCP methods with more sophisticated models are studied in enhanced compression model (ECM), which is a platform targeting at the next generation video coding standard, developed by JVET. This paper presents two methods, known as local-boosting CCP (LB-CCP) and non-local CCP (NL-CCP) to improve CCP with local and non-local information. With LB-CCP, prediction samples of CCP can be filtered with neighbouring samples. Besides, neighbouring template costs are calculated to determine the range of training samples, as well as the cross-component prediction method used in the chroma fusion mode. With NL-CCP, a CCP model can be derived with samples non-adjacent to the current block. As an equivalent but simpler implementation, the CCP model can be inherited from a CCP-coded neighbouring block as a spatial CCP candidate. Moreover, the CCP model of a previous CCP-coded block can be stored in a history-based table, which can be fetched and used by the current block as a history-based CCP candidate. A NL-CCP candidate list is built with the two kinds of candidates. The encoder can select the optimal candidate and send an index to the decoder. Experimental results show that LB-CCP together with NL-CCP can provide an average Bjontegaard delta rate (BD-rate) reduction of 0.27%, 2.31%, 2.44% on Y, Cb, Cr components, respectively, with a negligible change in the running time, compared with ECM-8.0 in all intra configurations. Both LB-CCP and NL-CCP have been adopted into ECM.

Lightweight Knowledge Distillation and Feature Compression Model for User Click-Through Rates Prediction in Edge Computing Scenarios

Lightweight Knowledge Distillation and Feature Compression Model for User Click-Through Rates Prediction in Edge Computing Scenarios

Preliminary design and analysis of an afterburner module

In this study, design calculations and calculations of afterburner used in jet engines are presented by evaluating the results of analytical and computational analysis. Afterburner inlet values of 1050 K temperature, 300 kPa pressure and 3.6 kg/s mass flow rate are taken as the design conditions. Maximum length and diameter are 500 mm and 200 mm, respectively, for the sections to be designed as length constraints within the scope of the conceptual design. The two-ring vee-gutter has 1.33 cm in diameter and 4.25 cm high. Jet A fuel is assumed to be injected into the core flow (90 degrees) from the spray bars. The spray is mounted in line with the vee-gutter to optimize the mixing of the flow. Analyses are performed for 4 cm between the spray bar and the vee-gutter. For the study, the GE J79 engine was examined from the literature and taken as a basis for the aerodynamic transition section design. Within the scope of TEKNOFEST 2023 Jet Engine Design Competition, a preliminary design of an afterburner module that can produce 700 pounds of thrust and has a life capacity of 25 hours should be realized in line with the design requirements and constraints. In the light of geometric constraints, one-dimensional combustion calculations of the module are made, and the parts are modelled using the relevant SolidWorks CAD program and these modelled parts are then transferred to ANSYS™ environment and the results and analyses are verified. The afterburner module flow analysis software program ANSYS™ is used to analyse the afterburner operation in both cold operating ranges (i.e. without combustion) using compressible, viscous and standard k-epsilon turbulence model. As a result, the effect of afterburner length on combustion performance is found to be significant. As a result of the calculations, afterburner length is found as 28.14 cm. It is found that the combustion efficiency is 81.5% and the temperature can be increased from 1050 K to 2044 K. The total pressure loss is 14.96% as pressure drop due to the geometric parameters and heat addition. The blockage ratio calculated due to the use of vee-gutter geometry is the most important parameter in the pressure drop. It is found that the jet engine producing 670 lbs (2981 N) of thrust when the afterburner is not working whereas it can provide a significant amount of power increment at the expense of a 50% increase in specific fuel consumption when the afterburner is active as well as providing a 738 lbs (3238 N) with a 10.1% thrust increase. The reason of the enhancement is mainly coming from the mixing effect of the vee-gutter on the flow structure.

Numerical study on dynamic responses and residual axial bearing capacity of square CFDST columns under lateral impact

This paper presented a numerical investigation into the dynamic responses and residual axial bearing capacity of square concrete-filled double-skin steel tube (CFDST) columns under lateral impact using ABAQUS. The finite element analysis (FEA) model for lateral impact and post-impact axial compression of square CFDST members was established, where the material nonlinearity and strain rate of steel tubes and concrete were considered and validated by existing experiments and studies. Based on the verified FEA model, the failure mechanism of the impact response process of the square CFDST member was revealed. To further explore the behavior of CFDST under axial compression after lateral impact, the typical axial load-axial displacement curves, axial strains of the steel tube, failure mode, and stress development of different components of the members were analyzed in depth. Parametric studies were conducted to clarify the influence of each parameter (including hollow ratio, slenderness ratio, concrete strength, and steel strength) on the residual capacity of post-impact axial compression in square CFDST members at different impact energies. Results demonstrated a linear correlation between residual axial bearing capacity coefficient and mid-span residual deflection after impact. Finally, a simplified formula for calculating the residual axial bearing capacity of square CFDST columns after lateral impact was proposed, which provided a reference for practical engineering applications.

Strength Prediction of Non-Sintered Hwangto-Substituted Concrete Using the Ultrasonic Velocity Method.

This paper presents and investigates the properties of concrete in which a portion of the cement is substituted with non-sintered Hwangto (NSH), a readily available building material in Asia. Given the inactive nature of NSH, this study aimed to determine the optimal cement replacement ratio and quantitative strength of the material. The unit weight, compressive strength, ultrasonic pulse velocity (UPV), and stress-strain of the NSH concrete (NSHC) were evaluated. Additionally, we developed a predictive model for determining compressive strength based on the regression analysis of compressive strength and UPV. The water-to-binder ratio was set to 0.41, 0.33, and 0.28, and the NSH replacement rates in the cement were set to 0%, 15%, 30%, and 45% for evaluating various strength ranges. The mechanical property measurements indicated reductions of 5.35% in unit weight, 35.62% in compressive strength, and 6.34% in UPV as the NSH was replaced. Notably, the smallest deviation from plain concrete was observed at a replacement rate of 15%. The scanning electron microscopy analysis results showed that the plain concrete exhibited a crystalloid structure; however, as the NSH replacement rate increased, the amorphous structure and pores increased while unreacted NSH particles were also observed. The X-ray diffraction analysis results demonstrate that the peak intensities for kaolinite and mullite increased as the NSH replacement rate increased, while those of C-S-H gel and CaO showed low peak intensities. Furthermore, the regression analysis concluded that an exponential function was suitable. Consequently, a compressive strength prediction model was developed, and in the error test, the NSHC model demonstrated an average error of <10%, with fewer errors at the lower compressive strength boundaries.

Ultrasonic and structural properties studies of niobium tungsten tellurite glasses

Utilizing the melt quenching procedure, new ternary tellurite glasses with the formula 80TeO2 - (20-X) WO3 - XNb2O5, where X = 0, 1, 3, 5, 7.5, and 10 mol%, were formed. X-ray diffraction has been used to study the glass samples' amorphous nature. Density was measured then molar volume was calculated to get more information about the prepared samples. The elastic moduli (longitudinal, bulk, shear, and young moduli) were derived from the ultrasonic velocities (longitudinal and shear) which have been obtained using the pulse-echo procedure. Also, experimental values of Poisson's ratio, and Debye temperature were achieved. According to the bond compression and Makishima-Mackenzie models, the elastic moduli were theoretically calculated, and the results were then compared to the experimental results. The average stretching force constant F‾, the glass network bond number per volume unit nb, the average crosslink density (n‾c), packing density Vi, and the bond energy per volume unit Em will all be used in the theoretical analysis of elastic moduli. DSC was used to measure the glass transition temperature (Tg) and the thermal stability (S) of the samples. FTIR was used to investigate the structure of the glass systems.

Dynamic behaviors of cavitation bubbles near biomimetic surfaces: A numerical study

This work is devoted to numerical studies of dynamic behaviors of underwater cavitation bubbles and the associated fluid flow phenomena such as high-speed liquid jet and pressure loads characteristics near biomimetic shark skin surfaces. Potential damages to the surface caused by cavitation bubbles are numerically quantified by the pressure peak, pressure duration and pressure impulse at the surface center. This work utilizes a compressible VOF multiphase model in conjunction with user defined functions (UDFs). The model has been validated against published experimental data in terms of bubble morphological evolution and the time-dependent liquid film thickness. The results reveal that the shark skin-inspired surfaces have important influence on bubble dynamics, water jets and shock wave pressures under various stand-off parameters. Violent deformation and splashing of the bubbles with weak liquid jet impact and pressure impulse can be observed on the surface covered with triangular shaped structures, accompanied by a certain degree of attenuation on the impacting strength and range of the shock waves. Additionally, the migrating and coalescing behaviors of bubble pairs are reduced with the increase of the initial interbubble distance. The large bubble–wall distance can considerably alter the direction of the microjet flow away from the shark skin-inspired surface. These findings shed the light for designing and controlling of multiple cavitation bubbles collapse near biomimetic surfaces of high-speed marine applications.

Hyper-tuning gene expression programming to develop interpretable prediction models for the strength of corncob ash-modified geopolymer concrete

Geopolymer concrete has been offered as an alternative to traditional concrete due to the environmental consequences of cement manufacture. It is also unable to make adjustments if the concrete's strength after casting does not meet the required strength. Because of this, it is highly beneficial to make strength predictions before pouring concrete. This study proposes applying the gene expression programming (GEP) method to forecast the mechanical strength of slag and corncob ash-based geopolymer concrete (SCA-GPC). The SHapley Additive exPlanations (SHAP) approach was used to identify the significance of parameters used for modeling. Best-fitted simulations and exceptional prediction performance were produced by the GEP (R2 = 0.96, MAE = 1.876 MPa, MAPE = 6.00%, RMSE = 2.417 MPa, and NSE = 0.954), (R2 = 0.94, MAE = 0.178 MPa, MAPE = 3.50%, RMSE = 0.236 MPa, and NSE = 0.942), and (R2 = 0.94, MAE = 0.126 MPa, MAPE = 3.60%, RMSE = 0.154 MPa, and NSE = 0.935) for compressive, flexural, and split tensile strengths models of SCA-GPC, respectively. Error and efficiency-based statistical checks confirmed the accuracy of the created models. The SHAP research also found that the mechanical strength of SCA-GPC was mostly controlled by slag quantity, curing age, water quantity, and fine aggregate quantity in the mix proportion. This research showed that the mechanical strength of SCA-GPC could be effectively predicted using the GEP method. The implementation of such methods will significantly improve the process of ensuring the quality of geopolymer concrete.

A compressed large language model embedding dataset of ICD 10 CM descriptions

This paper presents novel datasets providing numerical representations of ICD-10-CM codes by generating description embeddings using a large language model followed by a dimension reduction via autoencoder. The embeddings serve as informative input features for machine learning models by capturing relationships among categories and preserving inherent context information. The model generating the data was validated in two ways. First, the dimension reduction was validated using an autoencoder, and secondly, a supervised model was created to estimate the ICD-10-CM hierarchical categories. Results show that the dimension of the data can be reduced to as few as 10 dimensions while maintaining the ability to reproduce the original embeddings, with the fidelity decreasing as the reduced-dimension representation decreases. Multiple compression levels are provided, allowing users to choose as per their requirements, download and use without any other setup. The readily available datasets of ICD-10-CM codes are anticipated to be highly valuable for researchers in biomedical informatics, enabling more advanced analyses in the field. This approach has the potential to significantly improve the utility of ICD-10-CM codes in the biomedical domain.

Valproic acid ameliorates cauda equina injury by suppressing HDAC2-mediated ferroptosis.

Persistent neuroinflammatory response after cauda equina injury (CEI) lowers nociceptor firing thresholds, accompanied by pathological pain and decreasing extremity dysfunction. Histone deacetylation has been considered a key regulator of immunity, inflammation, and neurological dysfunction. Our previous study suggested that valproic acid (VPA), a histone deacetylase inhibitor, exhibited neuroprotective effects in rat models of CEI, although the underlying mechanism remains elusive. The cauda equina compression surgery was performed to establish the CEI model. The Basso, Beattie, Bresnahan score, and the von Frey filament test were carried out to measure the animal behavior. Immunofluorescence staining of myelin basic protein and GPX4 was carried out. In addition, transmission electron microscope analysis was used to assess the effect of VPA on the morphological changes of mitochondria. RNA-sequencing was conducted to clarify the underlying mechanism of VPA on CEI protection. In this current study, we revealed that the expression level of HDAC1 and HDAC2 was elevated after cauda equina compression model but was reversed by VPA treatment. Meanwhile, HDAC2 knockdown resulted in the improvement of motor functions and pathologic pain, similar to treatment with VPA. Histology analysis also showed that knockdown of histone deacetylase (HDAC)-2, but not HDAC1, remarkably alleviated cauda equina injury and demyelinating lesions. The potential mechanism may be related to lowering oxidative stress and inflammatory response in the injured region. Notably, the transcriptome sequencing indicated that the therapeutic effect of VPA may depend on HDAC2-mediated ferroptosis. Ferroptosis-related genes were analyzed invivo and DRG cells further validated the reliability of RNA-sequencing results, suggesting HDAC2-H4K12ac axis participated in epigenetic modulation of ferroptosis-related genes. HDAC2 is critically involved in the ferroptosis and neuroinflammation in cauda equina injury, and VPA ameliorated cauda equina injury by suppressing HDAC2-mediated ferroptosis.

Aged tendons lack adaptive response to acute compressive injury.

Rotator cuff tendinopathy has a multifactorial etiology, with both aging and external compression found to influence disease progression. However, tendon's response to these factors is still poorly understood and in vivo animal models make it difficult to decouple these effects. Therefore, we developed an explant culture model that allows us to directly apply compression to tendons and then observe their biological responses. Using this model, we applied a single acute compressive injury to C57BL/6J flexor digitorum longus tendon explants and observed changes in viability, metabolic activity, matrix composition, matrix biosynthesis, matrix structure, gene expression, and mechanical properties. We hypothesized that a single acute compressive load would result in an injury response in tendon and that this effect would be amplified in aged tendons. We found that young tendons had increased matrix turnover with a decrease in small leucine-rich proteoglycans, increase in compression-resistant proteoglycan aggrecan, increase in collagen synthesis, and an upregulation of collagen-degrading MMP-9. Aged tendons lacked any of these adaptive responses and instead had decreased metabolic activity and collagen synthesis. This implies that aged tendons lack the adaptation mechanisms required to return to homeostasis, and therefore are at greater risk for compression-induced injury. Overall, we present a novel compressive injury model that demonstrates lasting age-dependent changes and has the potential to examine the long-term response of tendon to a variety of compressive loading conditions.

DiffQuant: Reducing Compression Difference for Neural Network Quantization

Deep neural network quantization is a widely used method in the deployment of mobile or edge devices to effectively reduce memory overhead and speed up inference. However, quantization inevitably leads to a reduction in the performance and equivalence of models. Moreover, access to labeled datasets is often denied as they are considered valuable assets for companies or institutes. Consequently, performing quantization training becomes challenging without sufficient labeled datasets. To address these issues, we propose a novel quantization pipeline named DiffQuant, which can perform quantization training using unlabeled datasets. The pipeline includes two cores: the compression difference (CD) and model compression loss (MCL). The CD can measure the degree of equivalence loss between the full-precision and quantized models, and the MCL supports fine-tuning the quantized models using unlabeled data. In addition, we design a quantization training scheme that allows the quantization of both the batch normalization (BN) layer and the bias. Experimental results show that our method outperforms state-of-the-art methods on ResNet18/34/50 networks, maintaining performance with a reduced CD. We achieve Top-1 accuracies of 70.08%, 74.11%, and 76.16% on the ImageNet dataset for the 8-bit quantized ResNet18/34/50 models and reduce the gap to 0.55%, 0.61%, and 0.71% with the full-precision network, respectively. We achieve CD values of only 7.45%, 7.48%, and 8.52%, which allows DiffQuant to further exploit the potential of quantization.

Learned image compression via neighborhood-based attention optimization and context modeling with multi-scale guiding

In recent years, learned image compression has witnessed significant advancements. However, many existing learned image compression models predominantly rely on Convolutional Neural Networks (CNNs) and predicting the distribution of latent features to use ever more expensive entropy models, which cannot fully capture detail redundancy and lead to prohibitively slow decoding. To address this challenge, this paper introduces novel image compression models, MGIC and RMGIC, which consist of the Residual Neighborhood-based Attention Module (RNAM), Multi-Scale Guide Entropy Model (MGEM) and Refined Multi-Scale Guide Entropy Model (RMGEM). Specifically, Neighborhood Attention (NA) is combined with global feature learning to construct the RNAM, which can accurately capture the correlation of spatial neighborhoods and is more suitable for image compression. The proposed RNAM is flexible and could work as a plug-and-play component to enhance CNN-based models. Additionally, MGEM employs two mask convolutions with two kernel sizes to leverage multi-scale information for context modeling and eliminating the effect of parallel decoding on model performance. Building upon MGEM, RMGEM incorporates a refined two-stage spatial context modeling technique to enhance the utilization of spatial neighborhoods and achieve more precise entropy prediction. Furthermore, MGIC and RMGIC are derived by integrating RNAM, MGEM, and RMGEM into the variational autoencoder framework. The proposed methods can save more bandwidth and improve the quality of information storage. Extensive experimental evaluations have shown that the proposed method is effective and achieves state-of-the-art performance.

Baimai Ointment relieves chronic pain induced by chronic compression of dorsal root ganglion in rats by regulating neuroactive ligand-receptor interaction and HIF-1 signaling pathway

The Baimai Ointment with the effect of relaxing sinew and activating collaterals demonstrates a definite effect on Baimai disease with pain, spasm, stiffness and other symptoms, while the pharmacodynamic characteristics and mechanism of this agent remain unclear. In this study, a rat model of chronic compression of L4 dorsal root ganglion(CCD) was established by lumbar disc herniation, and the efficacy and mechanism of Baimai Ointment in the treatment of CCD were preliminarily explored by behavioral tests, side effect evaluation, network analysis, antagonist and molecular biology verification. The pharmacodynamic experiment indicated that Baimai Ointment significantly improved the pain thresholds(mechanical pain, thermal pain, and cold pain) and gait behavior of CCD model rats without causing tolerance or obvious toxic and side effects. Baimai Ointment inhibited the second-phase nociceptive response of mice in the formalin test, increased the hot plate threshold of normal mice, and down-regulated the expression of inflammatory cytokines in the spinal cord. Network analysis showed that Baimai Ointment had synergistic effect in the treatment of CCD and was related to descending inhibition/facilitation system and neuroinflammation. Furthermore, behavioral tests, Western blot, and immunofluorescence assay revealed that the pain-relieving effect of Baimai Ointment on CCD may be related to the regulation of the interaction between neuroactive ligand and receptors(neuroligands) such as CHRNA7, ADRA2A, and ADRB2, and the down-regulation of the expression of NOS2/pERK/PI3K, the core regulatory element of HIF-1 signaling pathway in spinal microglia. The findings preliminarily reveal the mechanism of relaxing sinew and activating collaterals of Baimai Ointment in the treatment of Baimai disease, providing a reference for the rational drug use and further research of this agent.

A mathematical model for parameter setting in discrete element numerical simulation

To rationalize the setting of joint parameters, model size, and initial value of vertical stress in simulation of mining of steeply inclined coal seams, a fault tree analysis method of discrete element numerical simulation was used and a mathematical model was proposed. A method of eliminating the influences of size-effect errors on the parameters of coal and rock samples was obtained based on previous work. Furthermore, the constitutive equation and eigenvalue determination formula of a joint discontinuity surface were established, and a method of determination of the joint parameters was proposed, forming the complete “coupling chain” between parameters for numerical simulation. In addition, a formula for the initial value of vertical stress was constructed by way of the compression and shear model of the element body. Also, the minimum dimension was determined by means of strength factor analysis of fracture mechanics. Taking the research literature as an example, the model size and initial value of vertical stress were calculated. On this basis, the physical parameters of coal samples, the physical parameters of coal rocks considering the influence of the size effect and the calculated coal rock joint parameters considering the influence of size effect were directly used to comparatively analyze the displacement and stress fields, thus verifying the reasonability and correctness of the mathematical model.

Optimisation and prediction of compressive properties for concrete containing recycled aggregates and rice husk ash using response surface methodology (RSM)

AbstractRice husk ash (RHA), an agricultural by‐product, has been added as supplementary cementitious material (SCM) in concrete mixture to improve the compressive properties of recycled aggregate concrete (RAC) in recent years. This study aimed to optimise the mixture design of RAC considering two variables: the replacement ratio (wt.%) of recycled aggregate (RA) to natural aggregate (NA) with three levels (0%, 50% and 100%) and the replacement ratio (wt.%) of RHA to cement with three levels (0%, 10% and 20%). Compression test was implemented at concrete age of 28 days based on the full factorial experiment. By means of response surface methodology (RSM), the optimised RA replacement ratio and RHA replacement ratio can be calculated with respect to the compressive strength and E‐modulus at 28 days, and vice versa the compressive strength and E‐modulus at 28 days of RAC containing RHA can also be predicted. According to response surface modelling, the compressive strength reaches the maximum value when the RA replacement ratio is 0% and the RHA replacement ratio is 7%, and the E‐modulus would reach the maximum when the RA replacement ratio is 17% and the RHA replacement ratio is 7%. The determination coefficient (R2) and adjusted coefficient (R2adj) for the compressive strength model are 0.9632 and 0.9544 respectively, and for the E‐modulus model are 0.9319 and 0.9157 respectively, showing that the models developed by RSM are relatively well correlated with the experimental results.

Effect of anisotropic interactions on the heat conduction of one-dimensional chains

One-dimensional (1D) chain models are frequently employed to investigate heat conduction in various systems. Despite their widespread use, there has been limited exploration of 1D chain models with anisotropic interactions. In this study, we examine heat conduction in a 1D chain model with orientation–position coupled interaction, namely the compressible XY model, and compare it with isotropic Fermi-Pasta-Ulam-Tsingou β (FPUT-β) systems. At low temperatures, the local temperatures in the translational and rotational degrees of freedom differ due to the difference in the contact thermal resistance in the two degrees of freedom. The system maintains orientational order, and the orientation–position-coupling effect lowers thermal conductivity in translational degrees of freedom. As the temperature rises above a transition point, the rotation of particles switches from oscillation to diffusion, leading to a solid-like to fluid-like transition in the rotational degree of freedom. The anisotropic interactions become negligible under time averaging, making the system’s properties similar to those of isotropic FPUT-β systems. Additionally, we investigate the system’s orientational structure to elucidate this transition. These findings will enhance our understanding of the dynamics of nanoscale anisotropic systems, such as the heat conduction of spin chains.

Existence and uniqueness of compressible MHD solutions for non-Newtonian fluids

In this paper, we study a compressible MHD model for one-dimensional non-Newtonian fluids. Strong nonlinearities are addressed using the consistent estimation method of approximate solutions. The existence and uniqueness of positive density local solutions are obtained under the compatibility condition.

General mechanisms for stabilizing weakly compressible multiphase models and their applications

General mechanisms for stabilizing weakly compressible multiphase models and their applications