Channel Expansion Research Articles

Hybrid channel expansion and squeeze network for hyperspectral image classification

Deep learning has been widely used in hyperspectral image (HSI) classification tasks and has achieved good results in various fields. However, the large gap between a large number of parameters to be adjusted and the limited labeled samples of HSI will greatly increase the risk of model overfitting and inevitably consume a large amount of time and memory. A lightweight hybrid channel expansion and squeeze network is proposed to solve the small sample set problem. First, a channel expansion and squeeze module (CESM) based on depthwise convolution is designed to obtain higher fine-grained features by increasing the width of network channels. At the same time, the introduction of efficient channel attention module can emphasize the importance of each channel and effectively improve the representation of CESM compression output. Second, multidimensional CESM is used to reconstruct the entire feature extraction process and further reduce the model complexity while maintaining the intrinsic physical characteristics of HSI. Finally, the extracted high-level abstract features are imported into the global average pooling classifier to obtain the prediction probability of each category. Experimental results on Indian Pines, Pavia University, and Salinas Scene datasets show that the accuracy of the proposed method can reach 98.66%, 99.61%, and 99.84%, respectively, under limited labeled samples, and the computational cost is significantly lower than other advanced methods.

A Study on the Channel Expansion VAE for Content-Based Image Retrieval

Content-based image retrieval (CBIR) focuses on video searching with fine-tuning of pre-trained off-the-shelf features. CBIR is an intuitive method for image retrieval, although it still requires labeled datasets for fine-tuning due to the inefficiency caused by annotation. Therefore, we explored an unsupervised model for feature extraction of image contents. We used a variational auto-encoder (VAE) expanding channel of neural networks and studied the activation of layer outputs. In this study, the channel expansion method boosted the capability of image retrieval by exploring more kernels and selecting a layer of comparatively activated object region. The experiment included a comparison of channel expansion and visualization of each layer in the encoder network. The proposed model achieved (52.7%) mAP, which outperformed (36.5%) the existing VAE on the MNIST dataset.

Effect of high-frequency beam oscillation on microstructures and cracks in laser cladding of Al-Cu-Mg alloys

This paper provides a new method to suppress solidification cracks in laser cladding of Al-Cu-Mg alloy by high-frequency beam oscillation without additional auxiliary measures. The effects of oscillation frequency on crack suppression were investigated by analyzing the microstructure evolution. The results show that the intercrystalline precipitated phase changed from coarse and continuous line to spiderweb-like morphology with the oscillation frequency (f) increased from 50 Hz to 500 Hz, and the thermal cracking tendency was completely reduced when the density of coarse intergranular precipitates was <300 μm/mm2 (f ≥ 200 Hz). The coarse Al2Cu and Al0.92Cu1.08Mg low-melting eutectic phases formed by the Cu element segregation between the columnar dendrites constituted the crack source and crack expansion channel. The rapid and periodic motion of a high-frequency oscillating laser can promote the uniform energy distribution and drive the melt flow, which will induce the columnar to equiaxed transition. The web-like eutectic liquid film between equiaxed dendrites has a smoother reflux channel and coordinated deformation ability. When the solidification shrinkage force pulls the local liquid film, the rest of the liquid phase can return to heal the crack in time, which is the main mechanism for the solidification cracks to be reduced.

Structural Optimization of a Muffler for a Marine Pumping System Based on Numerical Calculation

In order to eliminate the noise in the pumping system and prolong the service life of the equipment, an optimization design method for a muffler structure is proposed. The inside of the muffler is designed with a water guide cone which plays the role of diverting the fluid in the pipeline. The muffler shell, the wall surface of the water guide cone, and the circular bottom plate adopt a porous structure. In order to achieve the best sound suppression effect of the muffler structure, the effects of different channel expansion angles and the flow area on the performance of muffler are studied the results show that when the flow channel expansion angle of the muffler is 145° and the flow area ratio is 3 times, the noise elimination performance of the muffler is the best. The structural optimization of the muffler studied in this paper is of great significance to noise reduction in the pumping system.

Numerical study on detonation initiation process in the chamber with characteristic structures of the dual-mode scramjet combustor

In order to improve the performance of traditional ramjet, a ramjet based on pulse detonation (PD-Ramjet) instead of isobaric combustion was proposed. Several special-shaped detonation chambers which retained the characteristic structures of the dual-mode scramjet combustor were designed. Characteristic structures included the expansion channel and the cavity. The filling process and detonation initiation process of the stoichiometric hydrogen/air mixture under the incoming flow condition of the sub-combustion mode were studied in the detonation chambers with characteristic structures by two-dimensional numerical simulation method, and the influences of the characteristic structures on the filling process and detonation initiation process were analyzed. The simulation and analysis results indicated that the hydrogen concentration at both sides of the chamber near the outlet was low due to the structure of the expansion channel, and the pressure and velocity of the detonation wave decreased gradually after the expansion channel. The cavity had a significant influence on the filling process, which resulted in the uneven hydrogen concentration downstream the cavity, especially near the outlet of the chamber. The intensity of the detonation wave attenuated to a certain extent by reason of the cavity, where the peak pressure first decayed due to the sudden expansion of the flow path and then rose owing to the reflection at the cavity aft wall.

Numerical analysis of the spark channel expansion in a high-pressure hydrogen–oxygen mixture and in nitrogen

Numerical analysis of the spark channel expansion in a high-pressure hydrogen–oxygen mixture and in nitrogen

Newtonian hydro-thermal fluid fl 1 ow phenomena through a sudden 2 expansion channel with or without baffles

This work aims to study the different characteristics of Newtonian fluid flow and heat transfer througha 1:3 sudden expansion channel with or without plane baffles using the finite volume method. The flowis assumed viscous, incompressible, steady, and laminar. The different characteristics of hydro-thermalfluid flow phenomena have been studied for 𝑅𝑒 ∈ [0.1 − 200] to demonstrate the influence of thepresence of baffles. The profiles of velocity, pressure, skin friction coefficient, friction factor, averageNusselt number, and pumping power have been examined for both the cases of the presence and absenceof baffles. It has been observed that the hydro-thermal characteristics become more pronounced withthe increase in the number of baffles. In the case of three baffles of equal length, at 𝑅𝑒 = 200 it iscalculated that if the thickness of the baffles is same and equal to 10% of the length, then the value of Nu becomes approximately 1.2 times of that in the absence of baffle. For the same value of Re, ithas been found that for the presence of one baffle of length equal to the width of the inlet section of thechannel and of thickness 10% of the same, the value of 𝑁𝑢􀯔􀯩 becomes 1.11 times of that in the absenceof baffle, while in the case of three baffles of equal length and of equal thickness like the abovementioned case of one baffle, the value of 𝑁𝑢􀯔􀯩 becomes 1.25 times of that in absence of baffle. It hasalso been revealed that the enhancement of thermal phenomena increases with the increase inthe baffle’s height.

Turning on High‐Sensitive Organic Electrochemical Transistor‐Based Photoelectrochemical‐Type Sensor over Modulation of Fe‐MOF by PEDOT

AbstractLow‐gate voltages can make large modulations in the drain current. Accordingly, organic electrochemical transistors (OECTs) have attracted considerable interest as amplifying transducers. The use of OECTs in combination with electrochemical analysis promises to promote sensing performance. Herein, an OECT‐based photoelectrochemical‐type (OECT‐PEC) sensor is constructed based on the photo‐induced potential generating variation in the drain current. The as‐constructed OECT‐PEC sensor has high sensitivity because the device incorporated a sensor and an amplifier, providing the benefits of a PEC sensor and an OECT. In addition to the advantages of excellent signal amplification and ease of miniaturization, background signals are also effectively reduced, and possible interferences from electroactive substances are avoided due to the separation of light excitation and the electric output. The OECT‐PEC sensor integrates poly(3,4‐ethylenedioxythiophene) (PEDOT)‐modulated Fe‐metal organic framework (Fe‐MOF) nanocomposites as the photo‐active gating material and poly(3,4‐ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) as the channel material which exhibits a good photocurrent response from the gate electrode and a corresponding 100‐fold expansion channel response. With aptamer as the recognition element, a sensitive OECT‐PEC biosensor is constructed to detect the organophosphorus pesticide malathion. It is found to exhibit a good linear range, a low detection limit, and better sensitivity than the traditional PEC method.

ADDLight: An Energy-Saving Adder Neural Network for Cucumber Disease Classification

It is an urgent task to improve the applicability of the cucumber disease classification model in greenhouse edge-intelligent devices. The energy consumption of disease diagnosis models designed based on deep learning methods is a key factor affecting its applicability. Based on this motivation, two methods of reducing the model’s calculation amount and changing the calculation method of feature extraction were used in this study to reduce the model’s calculation energy consumption, thereby prolonging the working time of greenhouse edge devices deployed with disease models. First, a cucumber disease dataset with complex backgrounds is constructed in this study. Second, the random data enhancement method is used to enhance data during model training. Third, the conventional feature extraction module, depthwise separable feature extraction module, and the squeeze-and-excitation module are the main modules for constructing the classification model. In addition, the strategies of channel expansion and = shortcut connection are used to further improve the model’s classification accuracy. Finally, the additive feature extraction method is used to reconstruct the proposed model. The experimental results show that the computational energy consumption of the adder cucumber disease classification model is reduced by 96.1% compared with the convolutional neural network of the same structure. In addition, the model size is only 0.479 MB, the calculation amount is 0.03 GFLOPs, and the classification accuracy of cucumber disease images with complex backgrounds is 89.1%. All results prove that our model has high applicability in cucumber greenhouse intelligent equipment.

Systematic Literature Review on Remanufacturing Trade Based on Bibliometric Analysis

With the extensive development of remanufacturing, remanufacturing trade, as an essential part of it, has also attracted much attention from researchers. Quite a large number of studies related to remanufacturing trade, such as pricing, sales, competition, channel expansion and service strategy, have been published in various journals. This paper aims to focus on the research status on remanufacturing trade through bibliometric analysis that can provide the primary research trends and the future research hotspots by analyzing the progress, parties and themes of the research. In this paper, the review and analysis are conducted on over 121 articles from 2000 until July 2021 with the help of VOS viewer (Leiden University, Leiden, The Netherlands) and Citespace (Drexel University, Philadelphia, PA, USA). The results of the analysis of research progress and research parties suggest that: (a) more and more researchers have started to pay attention to consumers during modeling; (b) sustainable/green aspects are frequently taken into account as a factor, but the sustainable/green standards for remanufacturing trade need to be further developed; (c) studies on remanufacturing trade are more abundant in heavy industry countries. Additionally, the hotspots for future research are identified via the theme analysis, including: (1) with the issuing of various government policies, the impact of these policies on decisions related to remanufacturing trade has been studied gradually and is expected to become a hot topic in the future; (2) a large number of papers conduct research on the basis of the supply chain of remanufactured products; (3) research on the online distribution channel has already been carried out and is constantly deepening; (4) the digital technology is increasingly introduced to remanufacturing trade in recent years.

Continuous particle separation of microfluidic chip with integrated inertial separation and dielectrophoresis separation

Particle separation is essential in many microfluidic systems such as biomedical analysis and chemical reactions. This study aims to report a microfluidic separation device combining dielectrophoresis force and inertial force to separate particles continuously. Most particles were separated by inertial force via passing through the contraction and expansion channel and then deviated into different outlets via interdigital electrodes. Numerical simulations using the software COMSOL Multiphysics 5.4 were performed to investigate the effects of flow rate and electric field distribution on particles. The separation efficiency was assessed by separating 4 µm polystyrene spheres (PS) from 20 µm PS microspheres at various flow rates. The experimental results showed that the separation efficiency was more than 95%. This microfluidic chip is expected to be applied to cell sorting and biomedicine.

Using cloud-based virtual learning environments to mitigate increasing disparity in urban-rural academic competence

• Cloud-based Virtual Learning Environment (C-VLE) can help rural students to be academically competitive with their urban counterparts. • This study examined C-VLE adoption and its impact on rural students’ academic performance. • This study extended the task-technology fit theory with channel expansion theory and C-VLE related situational constructs. • This study can assist the educational authorities in increasing the quality of education in rural and remote areas. • This study has significant insights for researchers from technology management, public policy, and educational technology. The unequal distribution of educational opportunities between urban and rural areas is still a significant challenge in China. As a result, students from urban areas can benefit from highly qualified teachers and enjoy state-of-the-art facilities in contrast with their rural counterparts. Consequently, students from urban areas are more likely to join higher education institutes in comparison with their rural counterparts. In such a scenario, a cloud-based virtual learning environment (C-VLE) is of great importance to provide quality education in rural areas. This study aims to predict C-VLE usage's impact on academic performance and gender differences in rural China. The respondents belonged to rural areas and experienced varying levels of such difficulties. This study extended the task-technology fit theory with the channel expansion theory. The study revealed that task-technology fit, individual-technology fit, perceived media richness, omnipresence, and interactivity positively influence C-VLE usage that leads elevating academic performance. As a result, the students from rural areas have a better competitive position than their urban counterparts. Surprisingly, gender did not play any moderating role. The study provides significant theoretical and practical implications for scholars, educational authorities, and policymakers.

Efficient channel expansion and pyramid depthwise-pointwise-depthwise neural networks

Efficient channel expansion and pyramid depthwise-pointwise-depthwise neural networks

Hydrothermal analysis of water-Al2O3 nanofluid flow through a sudden expansion channel with intermediate step

This work presents the dynamic behavior of the laminar nanofluid hydrothermal flow through a sudden expansion channel with the variations in the normalized length, height, and pitch to width ratio of the intermediate step. The governing equations have been discretized by means of the finite volume method, and the SIMPLE algorithm has been applied to solve the system of algebraic equations. The effect of variations in the Reynolds number (Re), normalized height (B/C), normalized length (A/C), and step pitch-width ratio (A/B) on the enhancement of heat transfer have been studied. At various values of A/C, A/B and B/C, this work indicates that the average Nusselt number (Nuavg), thermal resistance factor (R), and performance number (PN) increase with the increase in Re. In the case of the hydrothermal flow phenomena, the configuration associated with B/C provides better results than the configurations associated with A/B and A/C. As an important outcome, we see that the presence of intermediate step enhances the average Nusselt number, and thermal resistance factor. This will become very much helpful for various realistic circumstances for triggering additional thermo-mechanical loads on the surface.

Multiple Prime Expansion Channel Hopping for Blind Rendezvous in a Wireless Sensor Network

A channel rendezvous is a significant aspect of communication. In this context, blind rendezvous is the process of selecting a common available channel and establishing a communication link for wireless devices in a wireless sensor network. The rendezvous of asymmetric and heterogeneous wireless devices is a challenge. Thus, to improve speeds and stability of rendezvous, we analyze time slot overlap and channel determinism in the rendezvous algorithm and propose a rendezvous algorithm named Multiple Prime Expansion (MPE). In a final simulation study, we compare the performance of the MPE with other existing algorithms in an asymmetric and heterogeneous scenario. Results show that MPE has excellent performance for the ATTR and MTTR.

Plasma blasting of rocks and rocks-like materials: An analytical model

Plasma blasting technology (PBT) is a potential alternative to chemical blasting and mechanical cutting methods for fragmentation of natural rocks, concrete, geopolymers , and other rocks-like materials. We present an analytical model of PBT addressing currently inadequate understanding of the dynamics of shock waves generation and propagation versus the electric energy release conditions. The proposed model describes the operation of the electrical discharge circuit, plasma channel initiation and expansion, and the generation and propagation of shock and pressure waves in the destructible solid. The dynamics of the power generator energy conversion into the plasma channel and into the wave of mechanical stresses in the solid are considered and the main factors determining the efficiency of the method, namely the pulse generator circuit parameters, exploding wire length, and shock wave-transmitting media, are evaluated. Solid fracture efficiency is shown to depend on the pressure pulse wave shape which, in turn, is determined by the rate of electrical energy deposition into the plasma channel. Increasing the exploding wire length leads to an earlier formation of the tensile tangential stresses and to their higher magnitude and thus facilitates material's fragmentation. The use of acoustically stiff media for shock wave transfer marginally improves material's fracture efficiency. Preliminary verification of the functionality of the model was carried out using commercial concretes, with good agreement between the analytically derived and experimentally obtained values. The results demonstrate that the proposed model allows to simulate PBT fracture over a wide range of instrumental and process conditions and can therefore be used for PBT process design, thus realising environmental and economic benefits through significant savings in time and experimental confirmation costs. • Electrical energy deposition rate determines plasma blasting technology efficiency. • Electro-fracture induces up to four zones of change of stresses and deformations. • Superposition of direct and reflected shock waves facilitates rocks fracture. • Increasing exploding-wire length increases shock wave energy and improves fracture. • Acoustically stiff media for shock-wave transfer improves fracture marginally.

Liquid limit of marine soft clay encountering desalination percolation process

The diffuse double layer (DDL) theory indicated that the liquid limit (LL) of clays decreases based on pore-water salinity during salinization tests conducted in a laboratory. However, LL also decreased based on pore-water salinity when authors conducted geological investigations in the regression depositional zone in Lianyungang city, contradicting the DDL theory. This contradictory finding is inferred due to the coupled effect of salinity leaching and erosion of the colloid during a desalination percolation process. To confirm this hypothesis, 3 model tests of distilled water percolation and seven accelerated percolation tests were designed in this study, and the soil/water characteristics after desalination percolation were determined. The results reveal that the pore-water salinity leaches are combined with the colloidal erosion during the desalination percolation process, leading to the gradual decrease in LL based on the reduction of salinity and suggesting that the dominant effect is the colloidal erosion. At the initial stage of the desalination process, the colloids easily erode from the clay matrix due to the weakened salinity inhibition on the colloid release, leading to the seepage channel expansion, and hence the increase in colloid dimension and concentration in leachates throughout percolation. Based on the further decrease in pore-water salinity, the colloids in the clay matrix continuously release and inversely clog the inner channels of the soils, decreasing the colloid dimension and concentration in leachates. When the percolation process is sufficiently long, all salinity would be leached out from the clay matrix, and only the erosion of the colloid would be identified.

CEAM-YOLOv7: Improved YOLOv7 Based on Channel Expansion and Attention Mechanism for Driver Distraction Behavior Detection

Driver distraction behavior is prone to induce traffic accidents. Therefore, it is necessary to detect them to caution drivers in time for traffic safety. In driver behavior recognition, the variety of behaviors and the diversity of the driving environment can have a certain effect on detection accuracy, and the information loss is severe in most existing methods. These make it challenging to improve the real-time accuracy of driver distraction behavior. In this paper, we propose an improved YOLOv7 based on the channel expansion and attention mechanism for driver distraction behavior detection, named CEAM-YOLOv7. The global attention mechanism (GAM) module focuses on the key information to improve accuracy. With the insertion of GAM into the Backbone and Head of YOLOv7, the global dimensional interaction features are scaled up to enable the network to extract key features. Furthermore, In the CEAM-YOLOv7 architecture, the convolution computation has been significantly simplified, which is conducive to increasing the detection speed. Combined with the Inversion and contrast limited adaptive histogram equalization (CLAHE) image enhancement algorithm, a channel expansion (CE) algorithm for data augmentation is presented to further optimize the detection effect of infrared (IR) images. On the driver distraction IR dataset of Hunan University of Science and Technology (HNUST) and Hunan University (HNU), the verification results show that CEAM-YOLOv7 achieved 20.26% higher <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">mAP</i> compared to the original YOLOv7 model, and the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">FPS</i> reaches 156, which illustrate that CEAM-YOLOv7 outperforms state-of-the-art methods in both accuracy and speed.

Task-Specific Optimization of Virtual Channel Linear Prediction-Based Speech Dereverberation Front-End for Far-Field Speaker Verification

Developing a single-microphone speech denoising or dereverberation front-end for robust automatic speaker verification (ASV) in noisy far-field speaking scenarios is challenging. To address this problem, we present a novel front-end design that involves a recently proposed extension of the weighted prediction error (WPE) speech dereverberation algorithm, the virtual acoustic channel expansion (VACE)-WPE. It is demonstrated experimentally in this study that unlike the conventional WPE algorithm, the VACE-WPE can be explicitly trained to cancel out both late reverberation and background noise. To build the front-end, the VACE-WPE is first independently (pre)trained to produce "noisy" dereverberated signals. Subsequently, given a pretrained speaker embedding model, the VACE-WPE is additionally fine-tuned within a task-specific optimization (TSO) framework, causing the speaker embedding extracted from the processed signal to be similar to that extracted from the "noise-free" target signal. Moreover, to extend the application of the proposed front-end to more general, unconstrained "in-the-wild" ASV scenarios beyond controlled far-field conditions, we propose a distortion regularization method for the VACE-WPE within the TSO framework. The effectiveness of the proposed approach is verified on both far-field and in-the-wild ASV benchmarks, demonstrating its superiority over fully neural front-ends and other TSO methods in various cases.

VACE-WPE: Virtual Acoustic Channel Expansion Based on Neural Networks for Weighted Prediction Error-Based Speech Dereverberation

Speech dereverberation is an important issue for many real-world speech processing applications. Among the techniques developed, the weighted prediction error (WPE) algorithm has been widely adopted and advanced over the last decade, which blindly cancels out the late reverberation component from the reverberant mixture of microphone signals. In this study, we extend the neural-network-based virtual acoustic channel expansion (VACE) framework for the WPE-based speech dereverberation, a variant of the WPE that we recently proposed to enable the use of dual-channel WPE algorithm in a single-microphone speech dereverberation scenario. Based on the previous study, some ablation studies are conducted regarding the constituents of the VACE-WPE in an offline processing scenario. These studies reveal the characteristics of the system, thereby simplifying the architecture and leading to the introduction of new strategies for training the neural network for the VACE. Experimental results demonstrate that VACE-WPE (our PyTorch implementation and pre-trained models are available from <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/dreadbird06/vace_wpe</uri> ) considerably outperforms its single-channel counterpart in simulated noisy reverberant environments in terms of objective speech quality and is superior to the single-channel WPE as well as several fully neural speech dereverberation methods when employed as the front-end for the far-field automatic speech recognizer.