Search

Renewable energy, such as wind power and photovoltaic power, has uncertain and intermittent characteristics and zero marginal cost characteristics. The traditional power market mechanism is difficult to adapt to the new power system with a high proportion of renewable energy, and the original market system needs to be reformed. This paper discusses the application of a VCG auction mechanism in the electricity market, proposes a two-stage VCG market-clearing model based on the VCG mechanism, including the day-ahead market and the real-time market, and discusses the nature of the VCG mechanism. In order to address the discrepancy between the actual output of stochastic generator sets in the real-time market and their pre-scheduled output in the day-ahead market due to prediction deviations, a method for calculating punitive costs is proposed. A reallocation method based on market entities' contributing factors to budget imbalance is proposed to address the issue of budget imbalance under the VCG mechanism, in order to achieve revenue and expenditure balance. Through an example, the incentive compatibility characteristics of the VCG mechanism are verified, the problems of the locational marginal pricing (LMP) mechanism in the stochastic electricity market with a high proportion of renewable energy are analyzed, the electricity prices of the LMP mechanism and the VCG mechanism under different renewable energy proportions are compared, and the redistribution of the budget imbalance of the VCG mechanism is analyzed. [ABSTRACT FROM AUTHOR]

The Gasi reservoir in the Qinghai oilfield is a typical high-temperature and high-salinity reservoir, with an average temperature and average salinity of 70.0 °C and 152,144 mg/L, respectively. For over 30 years since 1990, water flooding has been the primary method for enhancing oil recovery. Recently, the Gasi reservoir has turned into a mature oilfield. It possesses a high water cut of 76% and a high total recovery rate of 47%. However, the main developing enhanced oil recovery (EOR) technology for the development of the Gasi reservoir in the next stage is yet to be determined. Surfactant–polymer (SP) flooding, which can reduce the oil–water interfacial tension and increase the viscosity of the water phase, has been widely applied to low-temperature and low-salinity reservoirs across China in the past few decades, but it has rarely been applied to high-temperature and high-salinity reservoirs such as the Gasi reservoir. In this study, the feasibility of SP flooding for high-temperature and high-salinity reservoirs was established. Thanks to the novel surfactant and polymer products, an SP flooding formula with surfactants ZC-2/B2 and polymer BRH-325 was proposed for Gasi. The formula showed a low interfacial tension of 10−2 mN/m and a high viscosity of 18 MPa·s in simulated reservoir conditions. The oil displacement experiment demonstrated that this formula can enhance the oil recovery rate by 26.95% upon water flooding at 64.64%. This study provides a feasible EOR candidate technology for high-temperature and high-salinity reservoirs, as exemplified by the Qinghai Gasi reservoir. [ABSTRACT FROM AUTHOR]

This paper examines the construction and maintenance of Chinese Han Buddhist temples in modern China against the backdrop of societal transformation. Initially, it analyzes the profound impact of social changes since the mid-19th century on Buddhist monasteries, including political turmoil, economic development, and urbanization. Furthermore, the paper explores how temples were reconstructed and revitalized within this historical context, highlighting the monastic community's unwavering commitment to protecting the Dharma and ensuring its enduring presence. Additionally, this paper also explores the role of charismatic monks in enhancing the sanctity of temples and the influence of Buddhist institutional frameworks on the dynamics of state and society. The study employs a multifaceted analysis to understand the complex interplay between temple construction, economic development, and the cultural heritage of Buddhism in China. [ABSTRACT FROM AUTHOR]

The centrifugal force field in a hydrocyclone was affected by the concave-wall curvature radius R0, and the mechanism underlying droplet deformation was closely related to the mass transfer efficiency. Numerical simulation and experimental data were collected to reveal the deformation characteristics and mechanism of a single droplet crossing concave-wall jet. Normalized interfacial energy γ and stretching performance were provided to investigate the droplet deformation process. The results showed that the droplet was stretched along the streamwise direction and shrank along the spanwise direction in the concave-wall jet. The droplet interfacial energy and deformation were the largest when the droplet crossed the jet boundary at t = 0.20 s. The maximum γ value increased with the increase in R0 by 57.3% to 71.4%, and the distance between the droplet and concave wall increased with R0. The Q-criterion was exported to show the increase in the vortex strength with the decrease in R0 at the jet boundary. The pressure distribution inside the droplet showed that the pressure decreased as R0 increased, while the pressure difference increased along the streamwise and wall-normal directions. This study suggested that the droplet breakup was more difficult for a smaller R0, which was beneficial for liquid–liquid heterogeneous separation. [ABSTRACT FROM AUTHOR]

Under the combined impact of climatic, socioeconomic, and environmental factors, the vegetation NPP change process and its responses to drive factors in the sub-regions of Mainland China are not clear. This study analyzes the changing pattern of vegetation NPP in China from 2000 to 2022 from the perspective of zoning and clarifies its response mechanism to climate-human interaction based on the gravity center model, third-order partial correlation coefficient and geographical detector. The results showed that: (1) There was an overall decreasing trend of vegetation NPP in China from the southeast to the northwest; (2) The vegetation NPP gravity center in Northeast, Northwest, and North China migrated southwards, while that of Southwest, Central South, and East China showed northward migration.;(3) Human activities played a dominant role in zones with increasing vegetation NPP from 2000 to 2010, while climate change greatly contributed to the increase in vegetation NPP during 2011–2022; (4) Human activities, such as deforestation and overgrazing, in Northeast and North China should be reduced to prevent vegetation ecosystem degradation, and the negative impact of human activities should be reduced to maintain the growth of vegetation NPP. This study was conducted to support decision-making for the precise restoration of ecosystems. [ABSTRACT FROM AUTHOR]

Salary satisfaction (SS) perception by employees can be affected by psychological impacts from the workplace setting. Landscape attributes of green and blue spaces (GBS) may account for this effect, but relevant evidence is rarely verified. In this study, a total of 56 Chinese industrial parks were chosen as study sites, where employee satisfaction was assessed by rating facial expression scores (happy, sad, and neutral emotions) in photos obtained from social networks (Sina Weibo and Douyin). The structures of the GBSs were characterized remotely by largeness of size, height, and visible ratio of green view (GVI) in a 2 km radius buffer area around the workplace. Street view images from Baidu map were selected for estimating GVI using a pre-trained deep learning model and botanical experts evaluating woody plants' diversity. The results indicated that SS can be estimated with the maximum likelihood analysis model against the happy score, which ranged within 8.37–18.38 (average: 13.30 ± 2.32) thousand RMB. A regression model indicated SS was lowered by a larger green space area in agreement with a reduced happy score. Further, sad scores in highland areas with tall plants and a strong depression on the happy score was associated with a greater plant diversity. Interesting from this study, the designed apparent size of green space should be considered in green space construction near a workplace to prevent perceptional decline towards SS, while blue space is irrelevant in this relationship. Similarly, the diversity of woody plants should be planned to control its negative impact on the perception of positive emotions, with plant diversity beyond a comfortable level perhaps further decreasing SS. [ABSTRACT FROM AUTHOR]

The voltage inverters have the high-frequency switching characteristic, which will generate massive high-frequency harmonics in the motors. Especially, the inductance of the high-speed permanent magnet synchronous motor (HPMSM) is designed to be small, and the high-frequency harmonic content will be higher than that of ordinary motors, which adversely affects the system. To overcome this problem, this paper proposed a harmonic suppression method based on the LC filter and the adaptive notch filter for HPMSMs. The LC filter is connected between the inverter and the HPMSM to filter out high-frequency harmonics, however, at the cost of the system resonance. Therefore, the adaptive notch filter with the frequency tracking capability is designed to offset specific resonant peaks by constructing an antiresonant peak. The least mean square adaptive algorithm automatically adjusts the filter parameters according to the variation of the input signal to ensure accurate filtering in complex cases. Simulation and experiment results prove the practicability and effectiveness of the proposed scheme. The harmonic contents of HPMSM are significantly reduced, and the dynamic response performance of the control system is improved. [ABSTRACT FROM AUTHOR]

Given the challenges encountered by industrial cameras, such as the randomness of sensor components, scattering, and polarization caused by optical defects, environmental factors, and other variables, the resulting noise hinders image recognition and leads to errors in subsequent image processing. In this study, we propose a RAW image denoising method based on BlockwiseUNet. By enabling local feature extraction and fusion, this approach enhances the network's capability to capture and suppress noise across multiple scales. We conducted extensive experiments on the SIDD benchmark (Smartphone Image Denoising Dataset), and the PSNR/SSIM value reached 51.25/0.992, which exceeds the current mainstream denoising methods. Additionally, our method demonstrates robustness to different noise levels and exhibits good generalization performance across various datasets. Furthermore, our proposed approach also exhibits certain advantages on the DND benchmark(Darmstadt Noise Dataset). [ABSTRACT FROM AUTHOR]

Drawing on Foucault's theoretical framework of "space and power", this paper examines the discursive construction of "knowledge" in the context of Chinese Buddhist education. It traces the historical transformation of Chinese Buddhist education from the traditional "Sangha Forest"(the monastic community; 叢林 Conglin) style education to the Buddhist Academy, and analyzes how modern Buddhism reshaped its social image and function from a faith-based to a knowledge-based culture. Furthermore, this paper explores the reasons why modern Buddhism requires "knowledge" as a bridge between its worldly and transcendental dimensions, and the roles of elite laymen and monasteries as "Buddhist Institutes" in the new discursive practice. [ABSTRACT FROM AUTHOR]

Airborne electromechanical actuators (EMAs) play a key role in the flight control system, and their health condition has a considerable impact on the flight status and safety of aircraft. Considering the multi-scale feature of fault signals and the fault diagnosis reliability for EMAs under complex working conditions, a novel fault diagnosis method of multi-scale feature fusion convolutional neural network (MSFFCNN) is proposed. Leveraging the multiple different scales' learning structure and attention mechanism-based feature fusion, the fault-related information can be effectively captured and learned, thereby improving the recognition ability and diagnostic performance of the network. The proposed method was evaluated by experiments and compared with the other three fault-diagnosis algorithms. The results show that the proposed MSFFCNN approach has a better diagnostic performance compared with the state-of-the-art fault diagnosis methods, which demonstrates the effectiveness and superiority of the proposed method. [ABSTRACT FROM AUTHOR]

Urban forests have important functions to alleviate air pollution, adjust the regional climate, and promote visitors' mental health and well-being. Humans' perceptions towards both atmospheric particulate matter (PM) and meteorological factors can be a gauge for assessing the functional services of urban forests. The geographical locations of host cities experiencing urbanization could take part in effects on emotional states of forest visitors. In this study, a total of 6309 facial photos of urban forest visitors were obtained from social networks in 42 cities of mainland China. Happy, sad, and neutral emotions were rated as percent scores in 2020, when the same-day air pollutants, meteorological factors, and socioeconomic indicators were recorded. The positive emotional index (PEI) was calculated as the difference between happy scores and sad scores. The results reveal that severe air pollutants (jointly PM2.5 > 75 μg/m3, PM10 > 150 μg/m3, and AQI > 150) were more frequently found in cities in the northeastern and northern areas of China. The forest visitors in the northeastern cities showed higher happiness scores compared to the visitors in other regions. The Quasi-Poisson regression suggested that high scores of happiness were frequently disclosed in weathers with low PM10. High scores of sadness were regressed on exposure to cities with a low GDP per capita and low total retail sales but with a high GDP at low-longitudinal and high-latitudinal locations with low levels of PM2.5 pollution, relative humidity, and wind velocity and a high temperature. The happiness score and PEI showed high-value aggregations in mega-sized cities (population over 10 million), such as Beijing and Zhengzhou, and in a metro-sized city (population of 5–10 million in Langfang) from climate regions of China. [ABSTRACT FROM AUTHOR]

Most of the previous studies exploring the changing patterns of wetland in the Yellow River Delta (YRD) were conducted based on sparse time-series images, which ignored its severe environmental gradient and rapid evolution process of the wetland. The changes in the dominant factors in the evolution of the wetland in the YRD are not clear. This study used the dense time-series Sentinel-2 images to establish a wetland database of the YRD, and then analyzed the spatial distribution characteristics of, and temporal changes in, the wetland during 2015–2022. Finally, the dominant factors of the spatio-temporal evolutions of the wetland were explored and revealed. The results showed the following. (1) During 2015–2022, the wetland in the YRD was dominated by artificial wetland, accounting for 54.02% of the total wetland area in the study area. In 2015–2022, the total wetland area increased by 309.90 km2, including an increase of 222.63 km2 in natural wetlands and 87.27 km2 in artificial wetlands. In the conversion between wetland types, 218.73 km2 of artificial wetlands were converted into natural wetlands, and 75.18 km2 of natural wetlands were converted into artificial wetlands. The patch density of rivers, swamps, and salt pans increased, showing a trend of fragmentation. However, the overall degree of landscape fragmentation in wetlands weakened. The trend of changes in the number of patches and landscape shape index was the same, while the trend of changes in Shannon's diversity index and Contagion index was completely opposite. (2) Natural factors, such as precipitation (0.51, 2015; 0.65, 2016), DEM (0.57, 2017; 0.47, 2018; 0.49, 2020; 0.46, 2021), vegetation coverage (0.59, 2019), and temperature (0.48, 2022), were the dominant influencing factors of wetland changes in the YRD. The dominant single factor causing the changes in artificial wetlands was vegetation coverage, while socio-economic factors had lower explanatory power, with the average q value of 0.18. (3) During 2015–2022, the interactions between the natural and artificial factors of the wetland changes were mostly nonlinear and showed double-factor enhancement. The interactions between temperature and sunshine hours had the largest explanatory power for natural wetland change, while interactions between precipitation and vegetation coverage, and between temperature and vegetation coverage, had large contribution rates for artificial wetland change. The interactions among natural factors had the greatest impacts on wetland change, followed by interactions between natural factors and socio-economic factors, while interactions among socio-economic factors had more slight impacts on wetland change. The results can provide a scientific basis for regional wetland protection and management. [ABSTRACT FROM AUTHOR]

Developing renewable energy has become a major strategy for China to accelerate the energy transition and combat climate change. Accordingly, a guarantee mechanism for renewable energy consumption with renewable portfolio standards (RPS) has been set in China. However, currently, the top-down allocation of regional renewable energy consumption targets often has issues of unfairness and inefficiency. It is necessary to investigate the issue of how to stimulate the renewable energy consumption potential on the demand side and reasonably formulate the consumption responsibility weights of various market entities. This paper aimed to develop a new methodology for the weight allocation of renewable energy consumption responsibilities. In doing so, an integrated model of an evolutionary game and stochastic optimization was constructed between market entities and governments. Then, the equilibrium strategies of market entities and governments were obtained through the evolutionary game. Furthermore, based on the equilibrium strategies, this paper optimized the renewable energy consumption weight of each market entity, which constitutes the optimal allocation scheme of renewable energy consumption responsibility weights. Finally, using the data of 7069 market entities in Hubei Province in 2021, this study simulated the model to verify its effectiveness and practicability. The results indicate that the willingness of market entities to assume more consumption responsibility is positively correlated with the government's incentives and the maturity of the green electricity trading market. This study provides important implications for optimizing government regulations and promoting renewable energy consumption. [ABSTRACT FROM AUTHOR]

The aim was to explore the reserve of ecological and wellness tourism resources in the famous ecological and wellness base Hong Kong Wetland Park and further construct a systematic index system for the evaluation of ecological and wellness tourism resources in wetland parks. The DHGF algorithm was used to conduct hierarchical factor weighting analysis on ecotourism resource-related factors. The study showed that in terms of the index weights of ecological permaculture tourism resources in wetland parks, the highest weight was assigned to the permaculture tourism resources' own conditions. According to the evaluation system of eco-health tourism resources, the comprehensive evaluation score of eco-health tourism resources of Hong Kong Wetland Park is 7.673, and the comprehensive evaluation level is grade II (good). Based on the in-depth analysis of the evaluation results, it is proposed to establish a standardized government management system, explore ecological and health tourism resources with the concept of "original ecology and green health", cultivate "recreation and health tourism + ecological health" composite professionals in Hong Kong and Macao, and create the recommendations, including the establishment of a four-dimensional integration of "ecology + health + health + tourism" in the Greater Bay Area. [ABSTRACT FROM AUTHOR]

It is widely accepted that houses in better-designed neighborhoods are found to enjoy a price premium. Prior studies have mainly examined the impact of macro-level neighborhood attributes (e.g., park accessibility using land use data) on housing prices. More recently, research has investigated the micro-level features using street view imagery (SVI) data, though scholars limited the scope to objective indicators such as the green view index and sky view index. The role of subjectively measured street qualities is less discussed due to the lack of large-scale perception data. To provide better explanations of whether and how the micro-level neighborhood environment affects housing prices, this article introduces a framework to collect designers' perceptions on five subjective urban design perceptions from pairwise SVI rankings in Shanghai with an online visual survey and further predicted through machine learning (ML) algorithms. We also extracted ten important objective features from the scenes. The predictive power of micro-level neighborhood street perceptions (subjective perceptions and objective features) on housing prices was investigated using the hedonic price model (HPM) through ordinary least squares (OLS) and spatial regression, which considers spatial dependence. The findings prove the significance of the value of perceived qualities of the neighborhoods. It reveals that both objective perceived features and subjective perceptions significantly contribute to housing prices; while the objective features show more collective strengths, individual subjective perceptions have more explanatory power, and we argue that these two measures can complement each other. This study provides an important reference for decision makers when selecting street quality indicators to inform city planning, urban design, and community and housing development plans. [ABSTRACT FROM AUTHOR]

Deep-learning-based segmentation methods have achieved excellent results. As two main tasks in computer vision, instance segmentation and semantic segmentation are closely related and mutually beneficial. Spatial context information from the semantic features can also improve the accuracy of instance segmentation. Inspired by this, we propose a novel instance segmentation framework named partial atrous cascade R-CNN (PAC), which effectively improves the accuracy of the segmentation boundary. The proposed network innovates in two aspects: (1) A semantic branch with a partial atrous spatial pyramid extraction (PASPE) module is proposed in this paper. The module consists of atrous convolution layers with multi-dilation rates. By expanding the receptive field of the convolutional layer, multi-scale semantic features are greatly enriched. Experiments shows that the new branch obtains more accurate segmentation contours. (2) The proposed mask quality (MQ) module scores the intersection over union (IoU) between the predicted mask and the ground truth mask. Benefiting from the modified mask quality score, the quality of the segmentation results is judged credibly. Our proposed network is trained and tested on the MS COCO dataset. Compared with the benchmark, it brings consistent and noticeable improvements in the case of using the same backbone. [ABSTRACT FROM AUTHOR]

Endosymbionts living in plants and insects are pervasive. Ficus (Moraceae) has very special inflorescences (which we also call figs) enclosed like an urn, and such inflorescence is usually parasitized by fig wasps. Ficus breeds fig wasp larvae in its figs and adult fig wasps pollinate for Ficus, Ficus and its obligated pollinator formed fig-fig wasp mutualism. Previous studies have found that this confined environment in figs may have provided protection for fig wasps and that this has left some imprints on the genome of fig wasps during the coevolution history of figs and fig wasps. Research on the diversity of both bacteria and fungi in figs are fewer. Our study explored the diversity of endosymbionts in Ficus hirta figs. We utilized high-throughput sequencing and biological database to identify the specific microorganism in figs, then conducted microorganism communities' diversity analysis and function annotation analysis. As a result, we identified the dominant endosymbionts in figs, mainly some insect internal parasitic bacteria and fungi, plant pathogen, endophytes, and saprotroph. Then we also found bacteria in Ficus hirta figs were more diversified than fungi, and bacteria communities in female figs and functional male figs were different. These findings may give us more insight into the coevolution and interaction among endosymbiont, fig, and fig wasp. [ABSTRACT FROM AUTHOR]

Semantic segmentation, as a pixel-level recognition task, has been widely used in a variety of practical scenes. Most of the existing methods try to improve the performance of the network by fusing the information of high and low layers. This kind of simple concatenation or element-wise addition will lead to the problem of unbalanced fusion and low utilization of inter-level features. To solve this problem, we propose the Inter-Level Feature Balanced Fusion Network (IFBFNet) to guide the inter-level feature fusion towards a more balanced and effective direction. Our overall network architecture is based on the encoder–decoder architecture. In the encoder, we use a relatively deep convolution network to extract rich semantic information. In the decoder, skip-connections are added to connect and fuse low-level spatial features to restore a clearer boundary expression gradually. We add an inter-level feature balanced fusion module to each skip connection. Additionally, to better capture the boundary information, we added a shallower spatial information stream to supplement more spatial information details. Experiments have proved the effectiveness of our module. Our IFBFNet achieved a competitive performance on the Cityscapes dataset with only finely annotated data used for training and has been greatly improved on the baseline network. [ABSTRACT FROM AUTHOR]

Graph convolutional networks (GCNs), which model human actions as a series of spatial-temporal graphs, have recently achieved superior performance in skeleton-based action recognition. However, the existing methods mostly use the physical connections of joints to construct a spatial graph, resulting in limited topological information of the human skeleton. In addition, the action features in the time domain have not been fully explored. To better extract spatial-temporal features, we propose a multi-stage attention-enhanced sparse graph convolutional network (MS-ASGCN) for skeleton-based action recognition. To capture more abundant joint dependencies, we propose a new strategy for constructing skeleton graphs. This simulates bidirectional information flows between neighboring joints and pays greater attention to the information transmission between sparse joints. In addition, a part attention mechanism is proposed to learn the weight of each part and enhance the part-level feature learning. We introduce multiple streams of different stages and merge them in specific layers of the network to further improve the performance of the model. Our model is finally verified on two large-scale datasets, namely NTU-RGB+D and Skeleton-Kinetics. Experiments demonstrate that the proposed MS-ASGCN outperformed the previous state-of-the-art methods on both datasets. [ABSTRACT FROM AUTHOR]

3D object detection in LiDAR point clouds has been extensively used in autonomous driving, intelligent robotics, and augmented reality. Although the one-stage 3D detector has satisfactory training and inference speed, there are still some performance problems due to insufficient utilization of bird's eye view (BEV) information. In this paper, a new backbone network is proposed to complete the cross-layer fusion of multi-scale BEV feature maps, which makes full use of various information for detection. Specifically, our proposed backbone network can be divided into a coarse branch and a fine branch. In the coarse branch, we use the pyramidal feature hierarchy (PFH) to generate multi-scale BEV feature maps, which retain the advantages of different levels and serves as the input of the fine branch. In the fine branch, our proposed pyramid splitting and aggregation (PSA) module deeply integrates different levels of multi-scale feature maps, thereby improving the expressive ability of the final features. Extensive experiments on the challenging KITTI-3D benchmark show that our method has better performance in both 3D and BEV object detection compared with some previous state-of-the-art methods. Experimental results with average precision (AP) prove the effectiveness of our network. [ABSTRACT FROM AUTHOR]

Spatiotemporal graph convolution has made significant progress in skeleton-based action recognition in recent years. Most of the existing graph convolution methods take all the joints of the human skeleton as the overall modeling graph, ignoring the differences in the movement patterns of various parts of the human, and cannot well connect the relationship between the different parts of the human skeleton. To capture the unique features of different parts of human skeleton data and the correlation of different parts, we propose two new graph convolution methods: the whole graph convolution network (WGCN) and the part graph convolution network (PGCN). WGCN learns the whole scale skeleton spatiotemporal features according to the movement patterns and physical structure of the human skeleton. PGCN divides the human skeleton graph into several subgraphs to learn the part scale spatiotemporal features. Moreover, we propose an adaptive fusion module that combines the two features for multiple complementary adaptive fusion to obtain more effective skeleton features. By coupling these proposals, we build a whole and part adaptive fusion graph convolution neural network (WPGCN) that outperforms previous state-of-the-art methods on three large-scale datasets: NTU RGB+D 60, NTU RGB+D 120, and Kinetics Skeleton 400. [ABSTRACT FROM AUTHOR]

Three-dimensional object detection from point cloud data is becoming more and more significant, especially for autonomous driving applications. However, it is difficult for lidar to obtain the complete structure of an object in a real scene due to its scanning characteristics. Although the existing methods have made great progress, most of them ignore the prior information of object structure, such as symmetry. So, in this paper, we use the symmetry of the object to complete the missing part in the point cloud and then detect it. Specifically, we propose a two-stage detection framework. In the first stage, we adopt an encoder–decoder structure to generate the symmetry points of the foreground points and make the symmetry points and the non-empty voxel centers form an enhanced point cloud. In the second stage, the enhanced point cloud is input into the baseline, which is an anchor-based region proposal network, to generate the detection results. Extensive experiments on the challenging KITTI benchmark show the effectiveness of our method, which has better performance on both 3D and BEV (bird's eye view) object detection compared with some previous state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Deep neural networks are vulnerable to the adversarial example synthesized by adding imperceptible perturbations to the original image but can fool the classifier to provide wrong prediction outputs. This paper proposes an image restoration approach which provides a strong defense mechanism to provide robustness against adversarial attacks. We show that the unsupervised image restoration framework, deep image prior, can effectively eliminate the influence of adversarial perturbations. The proposed method uses multiple deep image prior networks called tandem deep image priors to recover the original image from adversarial example. Tandem deep image priors contain two deep image prior networks. The first network captures the main information of images and the second network recovers original image based on the prior information provided by the first network. The proposed method reduces the number of iterations originally required by deep image prior network and does not require adjusting the classifier or pre-training. It can be combined with other defensive methods. Our experiments show that the proposed method surprisingly achieves higher classification accuracy on ImageNet against a wide variety of adversarial attacks than previous state-of-the-art defense methods. [ABSTRACT FROM AUTHOR]

In this work, we present a convolutional neural network (CNN) named CGFA-CNN for blind image quality assessment (BIQA). A unique two-stage strategy is utilized which firstly identifies the distortion type in an image using Sub-Network I and then quantifies this distortion using Sub-Network II. Different from most deep neural networks, we extract hierarchical features as descriptors to enhance the image representation and design a feature aggregation layer in an end-to-end training manner applying Fisher encoding to visual vocabularies modeled by Gaussian mixture models (GMMs). Considering the authentic distortions and synthetic distortions, the hierarchical feature contains the characteristics of a CNN trained on the self-built dataset and a CNN trained on ImageNet. We evaluated our algorithm on four publicly available databases, and the results demonstrate that our CGFA-CNN has superior performance over other methods both on synthetic and authentic databases. [ABSTRACT FROM AUTHOR]

Deep learning based on a large number of high-quality data plays an important role in many industries. However, deep learning is hard to directly embed in the real-time system, because the data accumulation of the system depends on real-time acquisitions. However, the analysis tasks of such systems need to be carried out in real time, which makes it impossible to complete the analysis tasks by accumulating data for a long time. In order to solve the problems of high-quality data accumulation, high timeliness of the data analysis, and difficulty in embedding deep-learning algorithms directly in real-time systems, this paper proposes a new progressive deep-learning framework and conducts experiments on image recognition. The experimental results show that the proposed framework is effective and performs well and can reach a conclusion similar to the deep-learning framework based on large-scale data. [ABSTRACT FROM AUTHOR]

The purpose of this paper is to analyze the correlations among per capita gross domestic product (GDP), household fuel (natural gas and liquefied petroleum gas) consumption, and carbon dioxide (CO2) emissions through the environmental Kuznets curve (EKC) at the regional and national level in China using data from 2003 to 2015. The results validate the EKC assumption and show that per capita GDP is positively related to CO2 emissions; per capita natural gas consumption has a negative impact on CO2 emissions; however, per capita liquefied petroleum gas (LPG) consumption has a positive effect on CO2 emissions. Therefore, increasing natural gas consumption can effectively slow down the environmental degradation of China. Given rapid economic growth, changing the energy structure can improve the environment. [ABSTRACT FROM AUTHOR]

3D object detection in LiDAR point clouds has been extensively used in autonomous driving, intelligent robotics, and augmented reality. Although the one-stage 3D detector has satisfactory training and inference speed, there are still some performance problems due to insufficient utilization of bird's eye view (BEV) information. In this paper, a new backbone network is proposed to complete the cross-layer fusion of multi-scale BEV feature maps, which makes full use of various information for detection. Specifically, our proposed backbone network can be divided into a coarse branch and a fine branch. In the coarse branch, we use the pyramidal feature hierarchy (PFH) to generate multi-scale BEV feature maps, which retain the advantages of different levels and serves as the input of the fine branch. In the fine branch, our proposed pyramid splitting and aggregation (PSA) module deeply integrates different levels of multi-scale feature maps, thereby improving the expressive ability of the final features. Extensive experiments on the challenging KITTI-3D benchmark show that our method has better performance in both 3D and BEV object detection compared with some previous state-of-the-art methods. Experimental results with average precision (AP) prove the effectiveness of our network.