Vacancies engineering has sparked a huge interest in enhancing photocatalytic activity, but monovacancy simultaneously conducts as either electron or hole acceptor , redox reaction, worsening charge transfer and catal...
详细信息
Vacancies engineering has sparked a huge interest in enhancing photocatalytic activity, but monovacancy simultaneously conducts as either electron or hole acceptor , redox reaction, worsening charge transfer and catalytic performance. Here, the concept of electronic inversion has been proposed through the simultaneous introduction of surface oxygen and S vacancies in CdIn2S4 (OSv-CIS). Consequently, under mild conditions, the well-designed OSv-CIS-200 demonstrated a strong rate of N-benzylidenebenzylamine production (2972.07 mu mol g-1 h-1) coupled with Hydrogen peroxide (H2O2) synthesis (2362.33 mu mol g-1 h-1) (PIH), which is 12.4 times higher than that of CdIn2S4. Density functional theory (DFT) simulation and characterization studies demonstrate that oxygen is introduced into the lattice on the surface of the material, reversing the charge distribution of the S vacancy , enhancing the polarity of the total charge distribution. It not only provides a huge built-in electric field (BEF) for guiding the orientation of the charge transfer, but also acts as a long-distance active site to accelerate reaction and prevent H2O2 decomposition. Our work offers a straightforward connection between the atomic defect and intrinsic properties for designing high-efficiency materials.
Open data Processing Services are used to solve the bottleneck of big data storage and operation. At the same time, massive trajectory data is generated, and the basic information of users' spatio-temporal histori...
详细信息
Open data Processing Services are used to solve the bottleneck of big data storage and operation. At the same time, massive trajectory data is generated, and the basic information of users' spatio-temporal historical data is provided, including points of interest and movement patterns. Improving the availability of published trajectory statistics data without compromising user privacy is critical. Differential privacy technology is a standard technology to realize the secure release of trajectory statistics data. Several research efforts have focused on secure publication of trajectory statistics data in a central environment by adding noise to a trusted third-party server. However, this central approach is vulnerable to privacy breaches, where adversaries can access real data by locking down the third-party server. The local differential privacy, based on a distributed architecture, overcomes this form of attack by allowing users to scramble personal data records before they are sent to third-party server. However, the existing distributed privacy protection schemes still have the balance problem of poor availability of data when ensuring privacy, as well as the problem of excessive operation cost. Therefore, a local differential privacy mechanism based on water-filling for secure release of trajectory statistics data (WF-LDPSR) is proposed in this paper. Firstly, in order to protect user privacy individually, a user automatic personalized segmentation method is proposed to determine the effective user sensitivity level automatically. Secondly, a distributed privacy protection model based on local differential privacy is designed to resist the attacks on the third-party server. Finally, in order to achieve the optimal allocation of privacy budget, the water-filling theorem in the field of communication is introduced. An adaptive privacy budget allocation algorithm based on water-filling theorem is proposed to realize the adaptive privacy budget allocation. In addit
Stereo Image Super-Resolution (SSR) holds great promise in improving the quality of stereo images by exploiting the complementary information between left and right views. Most SSR methods primarily focus on the inter...
详细信息
Edge devices play an increasingly important role in the convolutional neural network(CNN)inference. However, the large computation and storage requirements are challenging for resource-and powerconstrained hardware. T...
详细信息
Edge devices play an increasingly important role in the convolutional neural network(CNN)inference. However, the large computation and storage requirements are challenging for resource-and powerconstrained hardware. These limitations might be overcome by exploring the following:(a)error tolerance via approximate computing, such as stochastic computing(SC);(b)data sparsity, including the weight and activation sparsity. Although SC can perform complex calculations with compact and simple arithmetic circuits,traditional SC-based accelerators suffer from the low reconfigurability and long bitstream, further making it difficult to benefit from the data sparsity. In this paper, we propose spatial-parallel stochastic computing(SPSC), which improves the spatial parallelism of the SC-based multiplier to the full extent while consuming fewer logic gates than the fixed-point implementation. Moreover, we present SPA, a highly reconfigurable SPSC-based sparse CNN accelerator with the proposed hybrid zero-skipping scheme(HZSS), to efficiently take advantage of different zero-skipping strategies for different types of layers. Comprehensive experiments show that SPA with up to 2477.6 Gops/W outperforms existing several binary-weight accelerators, SC-based accelerators, and the sparse CNN accelerator considering energy efficiency.
In the semantic segmentation of high-resolution remote sensing images, utilizing the normalized Digital Surface Model (nDSM) that provides height information as auxiliary data and fusing it with the visible image can ...
详细信息
ISBN:
(纸本)9789819784929;9789819784936
In the semantic segmentation of high-resolution remote sensing images, utilizing the normalized Digital Surface Model (nDSM) that provides height information as auxiliary data and fusing it with the visible image can improve the accuracy of segmentation. However, the better utilization of complementarity between different modal features has not been fully explored. In this work, we propose a new dual-branch and multi-stage Bimodal Fusion Rectification network (BFRNet), which is end-to-end trainable. It consists of three modules: Channel and Spatial Fusion Rectification (CSFR) module, Edge Fusion Refinement (EFR) module, and Multiscale Feature Fusion (MSFF) module. The CSFR module integrates and rectifies multimodal features in both channel and spatial dimensions, achieving sufficient interaction and fusion between multimodal features. The EFR module obtains better multiscale edge features than single modality through feature fusion based on bimodal interactive edge attention and spatial gate, which helps to alleviate the edge loss of ground objects in single modality. The MSFF module is used to upsample and fuse multiscale features from EFR and CSFR to generate the final semantic segmentation results. The experimental results on the two public datasets, Vaihingen and Potsdam, provided by ISPRS, showcase the comparative advantage of the proposed method over other research methods.
As one of the promising technologies of wideband beamforming, the anti-mainlobe interference wideband beamforming (AWB) algorithm can effectively suppress mainlobe distortion and sidelobe level rise, thereby improving...
详细信息
As one of the promising technologies of wideband beamforming, the anti-mainlobe interference wideband beamforming (AWB) algorithm can effectively suppress mainlobe distortion and sidelobe level rise, thereby improving the output signal to interference plus noise ratio (SINR) performance. Therefore, an AWB algorithm is proposed via a feature fusion convolutional neural network (FCNN) in this paper, named as AWB-FCNN algorithm. It can improve the beamforming performance and ensure the computational efficiency. For this algorithm, an AWB algorithm firstly is used to generate the network training label. Then, an FCNN model is constructed to predict beamforming weight vectors, which consists of a feature extraction module, a feature fusion module, and a weight vector prediction module. Specially, an atrous convolution layer is introduced into the feature extraction module to extract dense features, which be achieved by enlarging the receptive field without increasing the parameters of the network. Besides, the feature fusion module is used to reduce the irrelevant features such as mainlobe interference by fusing features at different scales. Finally, the well-trained FCNN model can rapidly and precisely output beamforming weight vectors. Simulation results show that the proposed algorithm has excellent interference suppression ability and high computational efficiency.
Neural network models face two highly destructive threats in real-world applications: membership inference attacks (MIAs) and adversarial attacks (AAs). One compromises the model's confidentiality, leading to memb...
详细信息
Genetic programming hyperheuristic (GPHH) has recently become a promising methodology for large-scale dynamic path planning (LDPP) since it can produce reusable heuristics rather than disposable solutions. However, in...
详细信息
Multi-aspect dense retrieval aims to incorporate aspect information (e.g., brand and category) into dual encoders to facilitate relevance matching. As an early and representative multi-aspect dense retriever, MADRAL l...
WiFi-based technology is appealing for indoor localization due to the widely deployed infrastructures. Recently, path separation solutions have been proposed to address the multipath effects and achieve decimeter-leve...
详细信息
暂无评论