In recent years, machine learning (ML) and deep learning (DL) have shown remarkable improvement in computer vision, natural language processing, stock prediction, forecasting, and audio processing to ...name a few. The size of the trained DL model is large for these complex tasks, which makes it difficult to deploy on resource-constrained devices. For instance, size of the pre-trained VGG16 model trained on the ImageNet dataset is more than 500 MB. Resource-constrained devices such as mobile phones and internet of things devices have limited memory and less computation power. For real-time applications, the trained models should be deployed on resource-constrained devices. Popular convolutional neural network models have millions of parameters that leads to increase in the size of the trained model. Hence, it becomes essential to compress and accelerate these models before deploying on resource-constrained devices while making the least compromise with the model accuracy. It is a challenging task to retain the same accuracy after compressing the model. To address this challenge, in the last couple of years many researchers have suggested different techniques for model compression and acceleration. In this paper, we have presented a survey of various techniques suggested for compressing and accelerating the ML and DL models. We have also discussed the challenges of the existing techniques and have provided future research directions in the field.
Objective. The purpose of this study was to compare the perioperative complications and clinical efficacy of patients with cervical spondylosis with spinal cord compression (CSWSCC) with or without ...MRI T2WIHS (T2-weighted image high signal) by means of propensity matching score grouping. Methods. We analyzed a single-center data of 913 surgical patients with CSWSCC by propensity matching score in this study, of which 326 patients had preoperative cervical MRI T2WIHS. The patient’s general condition and perioperative indicators were collected. The MRI T2WIHS and normal groups were paired 1 : 1 to eliminate selection bias by propensity matching score. Finally, a total of 312 pairs were matched successfully. The results of perioperative complications and other outcome variables were compared between the two groups by Cox function analysis. Results. The postoperative blood loss, operation time, blood transfusion volume, systemic complications, local complications, volume of drainage, abnormal use of antibiotic, length of hospital stay, and JOA (Japanese Orthopaedic Association) improvement rate were analyzed. As the only complication with significant statistical difference, the incidence of IRI (ischemia-reperfusion injury) in patients with MRI T2WIHS was significantly higher. The length of hospital stay was more significantly increased in patients with MRI T2WIHS; on the contrary, the JOA improvement rate decreased significantly. Conclusion. This study confirmed that there was no significant difference in the incidence of perioperative complications in CSWSCC patients with or without MRI T2WIHS, except for the IRI. Moreover, the JOA improvement rate of patients without MRI T2WIHS was significantly better, with the length of hospital stay reduced.
Due to concerns regarding the greenhouse effect and limitations on carbon dioxide emissions, the possibility of a next-generation combustion mode for internal combustion engines that can ...simultaneously reduce exhaust emissions and substantially improve thermal efficiency has drawn increasing attention. The most prominent characteristic of new combustion modes, such as Homogenous-Charge Compression-Ignition (HCCI), Stratified-Charge Compression-Ignition (SCCI), and Low-Temperature Combustion (LTC), is the requirement of creating a homogenous mixture or controllable stratified mixture prior to ignition. To this end, a lean fuel/air mixture and/or a controllable high level of exhaust gas recirculation (EGR) are employed to prolong the timescale of the ignition chemistry and port fuel injection or early in-cylinder injection is used to lengthen the mixing period. The mixture then undergoes controlled self-ignition near the top dead center (TDC) position due to the compression effect of the piston’s upward movement. It is worth noting that the entire combustion process lacks a direct method for the control of ignition timing and combustion rate, which are instead controlled primarily by chemical kinetics and, to a lesser extent, by turbulence and mixing. Because of the significant impacts of fuel physical–chemical properties on the ignition and combustion process, fuel design and management has become the most common approach for the control of ignition timing and combustion rate in such advanced combustion modes.
This paper summarizes the concepts and methods of fuel design and management and provides an overview of the effects of these strategies on ignition, combustion, and emissions for HCCI, LTC, and SCCI engines, respectively. From part 2 to part 4, the paper focuses on the effect of fuel design on HCCI combustion. A fuel index suitable for describing ignition characteristic under HCCI operating conditions is first introduced. Next, the proposed fuel design concept is described, including principles and main methodologies. Strategies based on the fuel design concept (including fuel additives, fuel blending, and dual-fuel technology) are discussed for primary reference fuels (PRF), alternative fuels, and practical gasoline and diesel fuels. Additionally, the effects of real-time fuel design on HCCI combustion fueled with PRFs and dimethyl ether/liquefied petroleum gas (DME–LPG) are evaluated. Diesel HCCI combustion has suffered from difficulties in homogenous mixture formation and an excessively high combustion rate. Therefore, LTC, which concentrates on local combustion temperature and a balance of mixture formation timescale and ignition timescale, has been proposed by many researchers. In Part 5, this paper provides an overview of the major points and research progress of LTC, with a preliminary discussion of the fundamental importance of fuel properties and fuel design strategy on the LTC process and emissions. Due to the stratification strategy has the capable of extending the HCCI operation range to higher loads, SCCI combustion, which incorporates HCCI combustion into a traditional combustion mode, has the potential to be used in commercial engines. Thus, this paper discusses the principles and control strategies of fuel design and management and also summarizes recent progress and future trends. The effect of fuel design and management on SCCI combustion is assessed for high cetane number fuels and high octane number fuels as well as the in SCCI combustion of gasoline–diesel dual-fuel and blends.
By leveraging data reduction technologies, such as data compression, all flash-based storage systems can have the same total cost of ownership (TCO) as traditional HDD-based storage systems. Thus, ...data compression has become a commodity feature for space efficiency and reliability in flash-based storage systems by reducing write traffic and space capacity demand. However, it introduces noticeable processing overheads on the critical I/O path, which degrades the system performance significantly. Existing data compression schemes for flash-based storage systems use fixed compression algorithms for all the incoming write data, failing to recognize and exploit the significant diversity in compressibility and access patterns of data and missing an opportunity to improve the system performance, the space efficiency or both. To achieve a reasonable trade-off between these two important design objectives, in this paper we introduce an Elastic Data Compression scheme, called EDC, which exploits the data compressibility and access intensity characteristics by judiciously matching data of different compressibility with different compression algorithms while leveraging the access idleness. Specifically, for compressible data blocks EDC exploits the compression diversity of the workload, and employs algorithms of higher compression rate in periods of lower system utilization and algorithms of lower compression rate in periods of higher system utilization. For non-compressible (or very lowly compressible) data blocks, it will write them through to the flash storage directly without any compression. The experiments conducted on our lightweight prototype implementation of the EDC system show that EDC saves storage space by up to 38.7 percent, with an average of 33.7 percent . In addition, it significantly outperforms the fixed compression schemes in the I/O performance measure by up to 61.4 percent, with an average of 36.7 percent.
This paper presents an experimental study of concrete at meso-scale (aggregates, macro-pores and mortar matrix) in order to get a better understanding of the local failure mechanisms known to drive ...the macroscopic mechanical behaviour of the material. The main originality comes from conducting in-situ X-ray mechanical tests on micro-concrete samples of realistic composition (including cement, sand, aggregates and water), under uniaxial compression and, for the first time, under triaxial compression at 5 MPa, 10 MPa and 15 MPa confining pressures. A timeseries analysis of the set of 3D images coming from each test allows for the measurement of the 3D kinematic fields (displacement and strain fields) throughout the experiments. The different failure patterns observed for each loading path are discussed, along with a quantification of the 3D fracturing processes at the scale of the largest heterogeneities (aggregates and macro-pores). With an increasing level of confinement, the transition from brittle to ductile response is observed, as well as an increase of the strength of the material. The pronounced impact of the meso-scale heterogeneities of concrete on their local failure mechanisms is highlighted. It is shown that strain localisation mainly originates between aggregates and mortar matrix, with the shape and location of the largest aggregates and macro-pores essentially driving the propagation of the cracking network.
Metastatic epidural spinal cord compression (MESCC) causes neurological deficits that may hinder ambulation. Understanding the prognostic factors associated with increased neurological recovery and ...regaining ambulatory functions is important for surgical planning in MESCC patients with neurological deficits. The present study was conducted to elucidate prognostic factors of neurological recovery in MESCC patients.
A total of 192 patients who had surgery for MESCC due to preoperative neurological deficits were reviewed. A motor recovery rate ≥ 50% and ambulatory function restoration were defined as the primary favorable endpoints. Factors associated with a motor recovery rate ≥ 50%, regaining ambulatory function, and patient survival were analyzed.
About one-half (48.4%) of the patients had a motor recovery rate ≥ 50%, and 24.4% of patients who were not able to walk due to MESCC before the surgery were able to walk after the operation. The factors "involvement of the thoracic spine" (p = 0.015) and "delayed operation" (p = 0.041) were associated with poor neurological recovery. Low preoperative muscle function grade was associated with a low likelihood of regaining ambulatory functions (p = 0.002). Furthermore, performing the operation ≥ 72 hours after the onset of the neurological deficit significantly decreased the likelihood of regaining ambulatory functions (p = 0.020). Postoperative ambulatory function significantly improved patient survival (p = 0.048).
Delayed operation and the involvement of the thoracic spine were poor prognostic factors for neurological recovery after MESCC surgery. Furthermore, a more severe preoperative neurological deficit was associated with a lesser likelihood of regaining ambulatory functions postoperatively. Earlier detection of motor weaknesses and expeditious surgical interventions are necessary, not only to improve patient functional status and quality of life but also to enhance survival.
The present study investigates torsional and compressive loading of a paperboard package. Finite element (FE) analyses simulating the tests were performed to improve understanding of the stresses and ...deformations in the paperboard during loading. A simple experimental characterization of the necessary material properties could be performed to represent the multi‐ply paperboard as a single‐ply structure. The results from the single‐ply model were compared with a laminate model, and the differences between the models were small. Comparing experimental and FE simulations of box compression and torsion showed that the FE models could accurately predict the response curves. However, in the simulations, there was an overprediction of the maximum compressive force and maximum torque, which was expected since geometrical imperfections and the heterogeneous internal structure of the material were not accounted for in the material model or the FE model. Local yield lines formed at the onset of non‐linearities in the package load–displacement curves. Therefore, the strength of the paperboard affects the maximum compressive strength and maximum torque, and the bending stiffness of the paperboard only had a minor effect. When a first local maximum was reached, the number of FE that reached the failure stress increased exponentially. The simulations also showed that box compression was not an effect of package height, but higher packages had a lower maximum torque.
Finite element simulations and physical experiments were performed to analyse torsional and compression loading of paperboard packages. The FE simulations accurately predicted the experimental load–deformation curves, with an expected overprediction, regardless of if the FE model was a single‐ply structure or a laminate model. Hence, the strength of the paperboard affects the maximum compressive strength and maximum torque, and the bending stiffness of the paperboard only had a minor effect.
To analyze the effect of different types of bone cement distribution after percutaneous vertebroplasty (PVP) in patients with osteoporotic vertebral compression fracture (OVCF).
One hundred thirty ...seven patients with single level OVCF who underwent PVP were retrospectively analyzed. The patients were divided into two groups according to bone cement distribution. Group A: bone cement contacted both upper and lower endplates; Group B: bone cement missed at least one endplate. Group B was divided into 3 subgroups. Group B1: bone cement only contacted the upper endplates; Group B2: bone cement only contacted the lower endplates; Group B3: bone cement only located in the middle of vertebral body. The visual analogue scale (VAS) score at 24 h post operation and last follow-up, anterior vertebral height restoration ratio (AVHRR), anterior vertebral height loss ratio (AVHLR), local kyphotic angle change and vertebral body recompression rate were compared.
24 h post operation, the pain of all groups were significantly improved. The average follow-up time was 15.3 ± 6.3 (6-24) months. At last follow-up, the VAS score of group A was lower than that of group B. There were 14 cases (10.2%) of adjacent vertebral fracture, 5 cases (8.6%) in group A and 9 cases (11.4%) in group B. There were 9 cases (6.6%) of cement leakage, 4 cases (6.9%) in group A and 5 cases (6.3%) in group B. At last follow-up, there were 16 cases (11.7%) of vertebral body recompression, including 3 cases (5.2%) in group A and 13 cases (16.5%) in group B. There was no significant difference in AVHRR between two groups. Local kyphotic angle change was significant larger in group B. At last follow-up, AVHLR in group B was higher than that in group A. Analysis in subgroup B revealed no significant difference in VAS score, local kyphotic angle change, vertebral recompression rate, AVHRR or AVHLR.
If the bone cement fully contacted both the upper and lower endplates, it can better restore the strength of the vertebral body and maintain the height of the vertebral body, reduce the risk of the vertebral body recompression and long-term pain.
Since the number of acquired images and their size have the tendency to increase, their lossy compression is widely applied for their storage, transfer, and dissemination. Simultaneously with ...providing a relatively large compression ratio, lossy compression produces distortions that are inevitably introduced and have to be controlled. The properties of these distortions depend on several factors such as image properties, the coder used, and a parameter that controls compression, which is different for particular coders. Then, one has to set a parameter that controls compression individually for an image to be compressed to provide image quality appropriate for a given application, and it is often desirable to do this quickly. Iterative procedures are usually not fast enough, and therefore fast and accurate procedures for providing a desired quality are needed. In the paper, such a procedure for two coders based on discrete cosine transform is proposed. This procedure is based on a prediction of mean square errors for a given quantization step using a simple analysis of image complexity (local activity in blocks). The statistical and spatial–spectral characteristics of distortions introduced by DCT-based coders are analyzed, and it is shown that they depend on the quantization step and local content. Generalizing the data for sets of grayscale test images and quantization step values, it is shown that the MSE can be easily predicted. These predictions are accurate enough and can be used to set the quantization step properly, as verified by experiments performed using more than 300 remote sensing and conventional optical images. The proposed approach is applicable to the lossy compression of grayscale images and the component-wise compression of multichannel data.
A series of incremental loading oedometer tests was conducted on pure clays and sand–clay mixtures with various sand/clay ratios and clay mineral compositions. The void indexes I
v
and I
vH
were ...introduced to evaluate their intrinsic compressibility. Test results revealed that I
v
was more suitable for depicting the compression behavior of pure clays than I
vH
; whereas, for the compressibility of sand–clay mixtures, the normalized compression line by using I
v
was obviously different from that of pure clays and traditional soils due to the presence of sand particles. Therefore, a four-phase analysis framework of sand–clay mixtures was introduced to unify the intrinsic compression behavior of soils with and without sands. It was found that the updated clay void index I
vc
parameter was effective to unify the soil’s compression behavior by excluding the influence of sand particles in clays before the sand skeleton formation. After the formation of the sand skeleton, the cluster particles change the stress distribution of mixtures, leading to less stress being applied on the clay matrix and thus the bifurcation of the intrinsic compression behavior.