Compressible Model Research Articles

Multilinear compressive learning and reconstruction of hyperspectral data with U-Net architecture for enhanced spectral imaging

Hyperspectral (HSI) data provide vast amounts of data which contain rich spatial and spectral information about each pixel of an image, such information is useful in detecting objects and identifying materials. However, the information provided by the HSI image is challenging to process. Hence, the compression of HSI makes it easier to utilize the data without the need for any heavy computation. Traditional compressive sensing techniques are adequate for lower dimensional signals, but they fail to preserve the essential spectral information provided by HSIs while performing compression. The proposed method aims to implement a Multilinear Compressive Learning (MCL) model for ideal compression of HSIs and a CNN based U-NET architecture is used for effective reconstruction of the compressed signal. The core of the algorithm involves two key blocks: (a) A multidimensional compressive sensing block for performing compression along every dimension, preserving spectral and spatial correlations. The employed sensing matrix is learned with respect to the given data for effective compression, (b) Reconstruction of the signal using a CNN based U-NET algorithm, which consists of encoders and decoders for effective reconstruction of the signal. Band selection is also performed in this technique to optimally select the most informative bands. The proposed model was implemented on the Indian Pines, Pavia University, and Salinas scenes datasets. The compression and reconstruction results were evaluated using performance metrics.

Syntax-Guided Content-Adaptive Transform for Image Compression.

The surge in image data has significantly increased the pressure on storage and transmission, posing new challenges for image compression technology. The structural texture of an image implies its statistical characteristics, which is effective for image encoding and decoding. Consequently, content-adaptive compression methods based on learning can better capture the content attributes of images, thereby enhancing encoding performance. However, learned image compression methods do not comprehensively account for both the global and local correlations among the pixels within an image. Moreover, they are constrained by rate-distortion optimization, which prevents the attainment of a compact representation of image attributes. To address these issues, we propose a syntax-guided content-adaptive transform framework that efficiently captures image attributes and enhances encoding efficiency. Firstly, we propose a syntax-refined side information module that fully leverages syntax and side information to guide the adaptive transformation of image attributes. Moreover, to more thoroughly exploit the global and local correlations in image space, we designed global-local modules, local-global modules, and upsampling/downsampling modules in codecs, further eliminating local and global redundancies. The experimental findings indicate that our proposed syntax-guided content-adaptive image compression model successfully adapts to the diverse complexities of different images, which enhances the efficiency of image compression. Concurrently, the method proposed has demonstrated outstanding performance across three benchmark datasets.

Effect of steam curing system on compressive strength of recycled aggregate concrete

Abstract Steam curing prefabrication is a feasible way to promote the application of recycled aggregate in the background of building industrialization. The combined use of slag and fly ash contributes to the comprehensive improvement of both the early and later-stage performance of recycled aggregate concrete (RAC). However, the steam curing regime for RAC composed of slag and fly ash remains unclear, including appropriate curing temperatures and time. In this study, the influences of curing temperatures (20, 40, 60, and 80°C), steam curing time (6, 9, and 12 h) and mineral admixtures (slag powder and fly ash) on the demoulding strength and strength development of RAC were investigated. Based on the Arrhenius formula, the hydration reaction rate was characterized by the development of compressive strength, the apparent activation energy related to the steam curing time was calculated, and the energy-based compressive strength model was proposed to realize the early strength prediction of the steam-cured RAC.

ADNNet: Attention-based deep neural network for Air Quality Index prediction

The Air Quality Index (AQI) is a crucial indicator for assessing the degree of atmospheric pollution. Accurately forecasting AQI is notably challenging due to the unpredictable weather conditions and the intricate interactions among various pollutants. To this end, we propose a AQI prediction framework, entitled ADNNet, to develop a streamlined attention-based method for AQI prediction. Specifically, it includes a deep neural network architecture that exclusively integrates multi-layer perceptron (MLP) and attention mechanism modules. This approach involves a novel design paradigm, where data information is first fed into the MLP layer to learn global features, which are then refined by the attention module to capture local features such as time delays and periodicity. Additionally, we incorporate a Bayesian hyperparameter optimization algorithm to make the trade-off between computational time and prediction accuracy efficient. To further enhance model performance, we apply exponential smoothing (ETS) and its inverse, named inverse-ETS, boosting the model’s predictive accuracy. We leverage air pollution datasets from various cities to verify the effectiveness of the proposed ADNNet. The results demonstrate that the proposed method significantly outperforms state-of-the-art (SOTA) models like LSTM, N-BEATS, Informer, Autoformer and VMD-TCN. The compressed model with reducing the mean absolute error (MAE) by 4.17%–16.12%, decreasing the root mean squared error (RMSE) by 3.52%–12.72%, and reducing the mean absolute scaled error (MASE) by 5.26%–17.28%.The source code of our model is available at https://github.com/xiankuiwu/AQI-Attention-based-DNN.

A New Insight into the Mechanical Behavior of Llzo through Experimental Pillar Compression and Molecular Dynamic Modeling

The transition to safe, high-energy-dense battery designs assumes the implementation of an all-solid-state lithium metal design. Solid electrolytes commonly short due to mechanical failure, where lithium dendrites will protrude into the surface until the anode and cathode are electrically connected. This phenomenon can be suppressed through materials engineering, i.e. designing solid electrolytes with specific mechanical properties. This work, for the first time, presents the measured compressive fracture strength of the oxide-based solid electrolyte. Fracture strength was measured using pillar compression with a flat punch tip and in-situ SEM imaging. Experimental validation of the measured fracture strength was completed using molecular dynamic modeling (MD), exploring the relationship of localized porosity to mechanical properties. Milled pillars (3 μm diameter) in the LLZO solid electrolyte show a linear relationship between the compressive yield strength and Young’s modulus, showing defects (porosity, cracks, voids, impurities, etc.) have a strong relationship on the ceramic’s physical response. MD modeling was used to introduce porosity to show the localized effect of porosity on the Young’s modulus and yield strength, providing strong evidence that localized porosity in the pillars may account for the substantial reduction in compressive yield strength. These insights shed light on the mechanical properties and electrochemical performance of LLZO, offering valuable guidance for the design of high-performance solid-state batteries. Figure 1

A Dynamic Damage Constitutive Model of Rock-like Materials Based on Elastic Tensile Strain

To accurately characterize the damage of rock-like materials under simultaneous or alternating tensile and compressive loading, a dynamic damage constitutive model for rock-like materials based on elastic tensile strain is developed by integrating the classical compressive plastic damage model and the tensile elastic damage model. The model is based on the Holmquist–Johnson–Cook (HJC) and Kuszmaul (KUS) models, categorizing the element stress state into tensile and compressive states through positive and negative elastic volumetric strain. It utilizes elastic tensile strain to enhance the calculation method for tensile cracks, determining the tensile strength of the principal direction based on the contribution rate of tensile principal stress for uniaxial/multiaxial loading. Additionally, it establishes a maximum elastic tensile strain rate function to rectify the model’s effect on the tensile strain rate. Through the LS-DYNA subroutine development, the model proficiently delineates the distribution of ring-shaped cracks on the frontal side and strip-shaped cracks on the rear side of the reinforced concrete slab subjected to impact loading. Numerical simulations demonstrate that the model provides more accurate damage prediction results for stress conditions involving simultaneous or alternating compression and tension, offering valuable insights for damage analysis in engineering blasting or impact penetration.

Experimental design, mechanical performance and mechanism analysis of C30 phosphogypsum-based concrete

The present study aimed to increase the utilization of phosphogypsum and reduce cement consumption. Single-factor experiments were designed to the effects of water-cement ratio (0.45–0.30), silica fume (1 %∼7 %), cement dosage (0 %∼20 %), and the RPG/HPG (20/80–0/100) on mechanical properties of phosphogypsum-based concretes (PGBCs). A new phosphogypsum-based concrete was prepared to meet the compressive strength requirement of C30 grade. The maximum 28d compressive strength of this PGBC reached 45.1 MPa. Simultaneously, the effect mechanisms of above four factors on the mechanical properties of PGBCs were discussed by combining XRD and SEM results. Moreover, the volume stability, elastic parameters, and compressive constitutive model of this PGBC with compressive strength above 30 MPa were investigated. This PGBC used more than 80 % phosphogypsum and 10∼20 % cement, exhibiting well economy and environmental friendliness. Moreover, the constitutive model was established to well predict the compressive behavior of PGBC with different RPG/HPG relative ratios. This PGBC had better volume stability and weaker deformation resistance than cement-based concrete. The present results can provide more theoretical and technical supports for the large utilization of PG.

Transplantation of miR-145a-5p modified M2 type microglia promotes the tissue repair of spinal cord injury in mice

BackgroundThe traumatic spinal cord injury (SCI) can cause immediate multi-faceted function loss or paralysis. Microglia, as one of tissue resident macrophages, has been reported to play a critical role in regulating inflammation response during SCI processes. And transplantation with M2 microglia into SCI mice promotes recovery of motor function. However, the M2 microglia can be easily re-educated and changed their phenotype due to the stimuli of tissue microenvironment. This study aimed to find a way to maintain the function of M2 microglia, which could exert an anti-inflammatory and pro-repair role, and further promote the repair of spinal cord injury.MethodsTo establish a standard murine spinal cord clip compression model using Dumont tying forceps. Using FACS, to sort microglia from C57BL/6 mice or CX3CR1GFP mice, and further culture them in vitro with different macrophage polarized medium. Also, to isolate primary microglia using density gradient centrifugation with the neonatal mice. To transfect miR-145a-5p into M2 microglia by Lipofectamine2000, and inject miR-145a-5p modified M2 microglia into the lesion sites of spinal cord for cell transplanted therapy. To evaluate the recovery of motor function in SCI mice through behavior analysis, immunofluorescence or histochemistry staining, Western blot and qRT-PCR detection. Application of reporter assay and molecular biology experiments to reveal the mechanism of miR-145a-5p modified M2 microglia therapy on SCI mice.ResultsWith in vitro experiments, we found that miR-145a-5p was highly expressed in M2 microglia, and miR-145a-5p overexpression could suppress M1 while promote M2 microglia polarization. And then delivery of miR-145a-5p overexpressed M2 microglia into the injured spinal cord area significantly accelerated locomotive recovery as well as prevented glia scar formation and neuron damage in mice, which was even better than M2 microglia transplantation. Further mechanisms showed that overexpressed miR-145a-5p in microglia inhibited the inflammatory response and maintained M2 macrophage phenotype by targeting TLR4/NF-κB signaling.ConclusionsThese findings indicate that transplantation of miR-145a-5p modified M2 microglia has more therapeutic potential for SCI than M2 microglia transplantation from epigenetic perspective.Graphical

A convergent finite-volume scheme for the diffusive compressible Euler model—resolved and under-resolved flows

The purpose of this paper is to demonstrate the robustness and accuracy of a provably convergent finite-volume scheme for the diffusive compressible Euler model. To this end, we consider sub- and supersonic flows as well as well-resolved smooth flows and under-resolved turbulent flows. The scheme is formulated for Voronoi meshes to allow complex geometries. It is completely defined without any parameters that require case-by-case tuning. Provable convergence guarantees both stability and admissibility of the numerical solutions, which is confirmed in our computations for both strong shocks and turbulence. For the turbulent Taylor-Green vortex, the scheme performs as well as a linearly-stable minimally-diffusive second-order scheme and for vortex shedding behind a cylinder, it even outperforms energy-stable second-order schemes.

Конечно-элементное моделирование нелинейных задач упругости в абсолютных узловых координатах на неструктурированных шестигранных сетках

We consider a finite element approach to solving the problem of elasticity theory in terms of absolute nodal coordinate formulation (ANCF), in which large body displacements are described in a global reference frame without using any local coordinate system. The main feature of the method is the absence of gyroscopic effects and, as a result, constancy of the mass matrix and the vector of the generalized force of gravity. In contrast to the traditional ANCF approach, the sets of nodal degrees of freedom of the finite element are formed only on the basis of absolute coordinates of nodes, which allows us to solve the problem using, in particular, unstructured hexahedral meshes. To construct the stiffness matrix, we apply a second-order automatic differentiation algorithm, which ensures its symmetrical form (Hessian matrix) and is analytically accurate in calculating the derivative. This approach also makes it possible to carry out calculations for models of hyperelastic materials without the corresponding Piola-Kirchhoff tensor. It has been shown that within the framework of discretization of the equation of motion, along with the well-known Newmark numerical integration scheme, it is possible to use the HTT-α scheme, which is unconditionally stable, second-order accurate and dissipative for high frequencies. We present several examples of solving static and dynamic elasticity problems for compressible and incompressible models of hyperelastic materials, in which the functions of internal energy density of the body are specified in terms of the deformation gradient.

On the Cauchy problem of 2D compressible fluid model with the horizontal thermal gradient effect

This article mainly investigates the existence of local strong solution to the Cauchy problem of the 2D compressible fluid model with the horizontal thermal gradient effect. By using delicate weighted energy estimates, the existence of a local strong solution is obtained with the initial density and the initial temperature decay not too slowly at infinity.

Application of Interactive Object Model in Sports Teaching

Traditional physical education is just repeated training without quantitative teaching and training data results. Multimedia information analysis and retrieval technology can make repeated research on the content of competition or training, and increase the quantitative analysis of training indicators. Therefore, this paper proposes an interactive compressed domain key video object model to realize the design and production of physical education video, so as to quantify the effect of physical education. This paper discusses the interactive design and application of PE teaching video from three aspects: static, dynamic, and semantic. The interactive object model proposed in this paper can formulate corresponding strategies for different key sports video objects. The distortion rate of the complete key video object is 2.3%, which is 0.09% lower than the distortion rate of the Kim2002 scheme. From a usage perspective, through the scheme proposed in this article, coaches can obtain teaching videos with simple initial extreme values on an interactive interface, and the distortion rate of the video object is lower than that of traditional schemes.

Intelligent Image Compression Model on the Basis of Wavelet Transform and Optimized Fuzzy C-Means-Based Vector Quantisation

Intelligent Image Compression Model on the Basis of Wavelet Transform and Optimized Fuzzy C-Means-Based Vector Quantisation

Investigating the damage characteristics of kiwifruits through the integration of compressive load and finite element method

Fruits are susceptible to various forms of mechanical damage after harvesting. Compression-induced bruising is the most common damage incurred, and it crucial to mitigate this for reducing economic losses associated with fruit. This study investigates the impact of compression on kiwifruit. Experiments combined with finite element method simulations were conducted to analyze the relationship between compression distance and bruise volume of kiwifruits under different storage times. This study's results found that compression distance and storage time are linearly positively correlated with the bruise volume of kiwifruit. The difference in bruising volume between simulation results and actual results is less than 14 %. The aforementioned results indicate that our employed compression model is capable of simulating real compression scenarios. This study provides significant theoretical insights into predicting and evaluating potential damages that may occur during post-harvest processing of kiwifruits under compression conditions.

Discretisations of mixed-dimensional Thermo-Hydro-Mechanical models preserving energy estimates

In this study, we explore mixed-dimensional Thermo-Hydro-Mechanical (THM) models in fractured porous media accounting for Coulomb frictional contact at matrix fracture interfaces. The simulation of such models plays an important role in many applications such as hydraulic stimulation in deep geothermal systems and assessing induced seismic risks in CO2 storage. We first extend to the mixed-dimensional framework the thermodynamically consistent THM models derived in [19] based on first and second principles of thermodynamics. Two formulations of the energy equation will be considered based either on energy conservation or on the entropy balance, assuming a vanishing thermo-poro-elastic dissipation. Our focus is on space time discretisations preserving energy estimates for both types of formulations and for a general single phase fluid thermodynamical model. This is achieved by a Finite Volume discretisation of the non-isothermal flow based on coercive fluxes and a tailored discretisation of the non-conservative convective terms. It is combined with a mixed Finite Element formulation of the contact-mechanical model with face-wise constant Lagrange multipliers accounting for the surface tractions, which preserves the dissipative properties of the contact terms. The discretisations of both THM formulations are investigated and compared in terms of convergence, accuracy and robustness on 2D test cases. It includes a Discrete Fracture Matrix model with a convection dominated thermal regime, and either a weakly compressible liquid or a highly compressible gas thermodynamical model.

Evaluating single event upsets in deep neural networks for semantic segmentation: An embedded system perspective

As the deployment of artificial intelligence (AI) algorithms at edge devices becomes increasingly prevalent, enhancing the robustness and reliability of autonomous AI-based perception and decision systems is becoming as relevant as precision and performance, especially in applications areas considered safety-critical such as autonomous driving and aerospace. This paper delves into the robustness assessment in embedded Deep Neural Networks (DNNs), particularly focusing on the impact of parameter perturbations produced by single event upsets (SEUs) on convolutional neural networks (CNN) for image semantic segmentation. By scrutinizing the layer-by-layer and bit-by-bit sensitivity of various encoder–decoder models to soft errors, this study thoroughly investigates the vulnerability of segmentation DNNs to SEUs and evaluates the consequences of techniques like model pruning and parameter quantization on the robustness of compressed models aimed at embedded implementations. The findings offer valuable insights into the mechanisms underlying SEU-induced failures that allow for evaluating the robustness of DNNs once trained in advance. Moreover, based on the collected data, we propose a set of practical lightweight error mitigation techniques with no memory or computational cost suitable for resource-constrained deployments. The code used to perform the fault injection (FI) campaign is available at https://github.com/jonGuti13/TensorFI2, while the code to implement proposed techniques is available at https://github.com/jonGuti13/parameterProtection.

Smartformer: An intelligent transformer compression framework for time-series modeling

Transformer, as one of the cutting-edge deep neural networks (DNNs), has achieved outstanding performance in time-series data analysis. However, this model usually requires large numbers of parameters to fit. Over-parameterization not only brings storage challenges in a resource-limited setting, but also inevitably results in the model over-fitting. Even though literature works introduced several ways to reduce the parameter size of Transformers, none of them addressed this over-parameterized issue by concurrently achieving the following three objectives: preserving the model architecture, maintaining the model performance, and reducing the model complexity (number of parameters). In this study, we propose an intelligent model compression framework, Smartformer, by incorporating reinforcement learning and CP-decomposition techniques to satisfy the aforementioned three objectives. In the experiment, we apply Smartformer and five baseline methods to two existing time-series Transformer models for model compression. The results demonstrate that our proposed Smartformer is the only method that consistently generates the compressed model on various scenarios by satisfying the three objectives. In particular, the Smartformer can mitigate the overfitting issue and thus improve the accuracy of the existing time-series models in all scenarios.

Decay of higher order derivatives for Lp solutions to the compressible fluid model of Korteweg type

As a follow-up study of [31], we present a derivation of upper bounds for the decay of arbitrary higher order derivatives of solutions to the compressible Navier-Stokes-Korteweg system in the Lp critical Besov space. The approach, based on Gevrey regularizing estimates, is to establish uniform bounds on the growth of the radius of analyticity of the solution in negative Besov norms, which may be applicable to a wide range of dissipative systems.

Study on Key Properties and Model Establishment of Innovative Recycled Aggregate Pervious Concrete.

In order to meet the needs of low-impact development and sustainable development, there is an urgent desire to develop an innovative recycled aggregate pervious concrete (I-RAPC) that is of high strength and permeability. In this study, I-RAPC was prepared based on response surface methodology (RSM) using recycled aggregate, river sand, and different types of pipes as the materials, and the effects of different pipe parameters (number, diameter, material, and distribution form) on the performance of I-RAPC were investigated. In addition, the calculation model of the compressive strength and the permeability coefficient of I-RAPC were proposed. The results showed that the frontal- and lateral-compressive strengths of I-RAPC were 39.8 MPa and 42.5 MPa, respectively, when the pipe material was acrylic, the position was 1EM, and the diameter was 10 mm-at which time the permeability coefficient was 3.02 mm/s, which was the highest in this study. The maximum relative errors of the compressive strength calculation model and the permeability coefficient calculation model were only 7.52% and 4.42%, respectively, as shown by the post hoc test. Therefore, I-RAPC has the advantages of high strength and permeability and is expected to be applied in low-impact development in cities with heavy surface sediment content and rainfall.

Reconstruction-free Image Compression for Machine Vision via Knowledge Transfer

Reconstruction-free image compression for machine vision aims to perform machine vision tasks directly on compressed-domain representations instead of reconstructed images. Existing reports have validated the feasibility of compressed-domain machine vision. However, we observe that when using recent learned compression models, the performance gap between compressed-domain and pixel-domain vision tasks is still large due to the lack of some natural inductive biases in pixel-domain convolutional neural networks. In this paper, we attempt to address this problem by transferring knowledge from pixel domain to compressed domain. A knowledge transfer loss defined at both output level and feature level is proposed to narrow the gap between compressed domain and pixel domain. In addition, we modify neural networks for pixel-domain vision tasks to better suit compressed-domain inputs. Experimental results on several machine vision tasks show that the proposed method improves the accuracy of compressed-domain vision tasks significantly, which even outperforms learning on reconstructed images while avoiding the computational cost of image reconstruction.