Search

Abstract Session-based recommendation is an important part of many e-commerce websites. Its purpose is to make recommendations based on the interaction behavior of anonymous users in a short period of time. Graph neural network can capture complex interactions in sessions, and they are a class of methods with better performance among existing session-based recommendation models. However, most existing models only learn item features for a single session. At the same time, GNNs are not good at capturing long-distance dependencies in a session, which leads to limited performance improvements for them. To address this deficiency, we propose sequence-aware graph neural network incorporating neighborhood information, named SAN-GNN. We construct a session graph and a neighborhood graph to learn item representations. For neighborhood graph, we propose a neighborhood Information extractor for the neighborhood graph to learn the neighbor information of nodes on the neighborhood graph. For the graph model of the session layer, we propose a session graph attention(SGA) module to learn the item representation of the target session. SGA uses Ta-LSTM to learn sequential dependencies in the target session and uses GCN with an integrated attention mechanism to learn node feature relevant to the target item. Exhaustive experiments on three public real-world datasets show that SAN-GNN outperforms the most advanced existing session-based method.

Abstract Named entity recognition and relation extraction are two important fundamental tasks in natural language processing. The joint entity-relationship extraction model based on parameter sharing can effectively reduce the impact of cascading errors on model performance by performing joint learning of entities and relationships in a single model, but it still cannot essentially get rid of the influence of pipeline models and suffers from entity information redundancy and inability to recognize overlapping entities. To this end, we propose a joint extraction model based on the decomposition strategy of pointer mechanism is proposed. The joint extraction task is divided into two parts. First, identify the head entity, utilizing the positive gain effect of the head entity on tail entity identification.Then, utilize a hierarchical model to improve the accuracy of the tail entity and relationship identification. Meanwhile, we introduce a pointer model to obtain the joint features of entity boundaries and relationship types to achieve boundary-aware classification. The experimental results show that the model achieves better results on both NYT and WebNLG datasets.

Although progress has been made in multisource data scene parsing of natural scene images, extracting complex backgrounds from aerial images of various types and presenting the image at different scales remain challenging. Various factors in high-resolution aerial images (HRAIs), such as imaging blur, background clutter, object shadow, and high resolution, substantially reduce the integrity and accuracy of object segmentation. By applying multisource data fusion, as in scene parsing of natural scene images, we can solve the aforementioned problems through the integration of auxiliary data into HRAIs. To this end, we propose a multiscale cross-layer interactive and similarity refinement network (MISNet) for scene parsing of HRAIs. First, in a feature fusion optimization module, we extract, filter, and optimize multisource features and further guide and optimize the features using a feature guidance module. Second, a multiscale context aggregation module increases the receptive field, captures semantic information, and extracts rich multiscale background features. Third, a dense decoding module fuses the global guidance information and high-level fused features. We also propose a joint learning method based on feature similarity and a joint learning module to obtain deep multilevel information, enhance feature generation, and fuse multiscale and global features to enhance network representation for accurate scene parsing of HRAIs. Comprehensive experiments on two benchmark HRAIs datasets indicate that our proposed MISNet is qualitatively and quantitatively superior to similar state-of-the-art models.

The detection of defects is indispensable in industrial production. Surface defects have different scales. Both minimal flaws and significant scratches may appear on the same product. The standard method uses a multi-scale feature fusion network, introducing many parameters that may reduce the inference speed. In actual industrial production scenarios, inference speed and accuracy play an equally important role. Therefore we propose an algorithm to effectively improve the detection speed while improving the detection accuracy. The model proposed in this paper called “YOLO with lightweight feature fusion network (LFF-YOLO).” First, we use ShuffleNetv2 as a feature extraction network to reduce the number of parameters. Then, to improve the efficiency of multi-scale feature fusion, we propose the lightweight feature pyramid network (LFPN). Considering that the fixed receptive field is difficult to adapt to the defects of different scales, it may lead to the difficulty of model convergence and seriously affect the detection performance. Therefore, we propose the adaptive receptive field feature extraction (ARFFE) module, which weights the multi-receptive field channels to generate multi-receptive field information. In addition, focal loss is used to solve the problem of imbalance between positive and negative samples. Finally, we conducted experiments on NEU-DET (79.23% mAP), Peking University printed circuit board defect dataset (93.31% mAP),and GC10-DET (59.78% mAP), respectively. Extensive experiments show that our proposed method achieves optimal detection speed compared with the prevailing methods, and the detection accuracy of our method is also highly competitive. We open-soure our code in the following URL:https://github.com/syyang2022/LFF-YOLO

In fine-grained visual classification(FGVC), small inter-class variations and large intra-class variations are always inherent attributes, so it is much more challenging than traditional classification tasks. Recent studies have mainly tackled this problem by employing attention mechanisms to locate the most discriminative parts. However, these methods tend to neglect other inconspicuous but distinguishable parts, and can not effectively fuse the features information of different scales and different degrees of discrimination. In this paper, we propose a multi-scale progressive complementary fusion network (MPCF-Net) to address these problems. In particular, we propose the following: (i) A three-step multi-scale progressive training method, which employs an image slicer to generate puzzle images at different scales followed by multi-step progressive training. This enables the network to capture multi-granularity local feature information and gradually expand its attention to global structural information as the training progresses for multi-granularity information fusion. (ii) A plug-and-play feature complementary enhancement module (FCEM) that explicitly enhances the features extracted by the current layer of the network, while also enabling the next layer of the network to extract potential complementary feature information to diversify the features. Our experiments were conducted on four FGVC benchmark datasets and yielded state-of-the-art and competitive results.

Most existing recommendation models only consider single user–item interaction information, which leads to serious cold-start or data sparsity problems. In practical applications, a user’s behavior is multi-type, and different types of user behavior show different semantic information. To achieve more accurate recommendations, a major challenge comes from being able to handle heterogeneous behavior data from users more finely. To address this problem, this paper proposes a multi-behavior recommendation framework based on a graph neural network, which captures personalized semantics of specific behavior and thus distinguishes the importance of different behaviors for predicting the target behavior. Meanwhile, this model establishes dependency relationships among different types of interaction behaviors under the graph-based information transfer network, and the graph convolutional network is further used to capture the high-order complexity of interaction graphs. The experimental results of three benchmark datasets show that the proposed graph-based multi-behavior recommendation model displays significant improvements in recommendation accuracy compared to the baseline method.

Although significant progress has been made in scene classification of high-resolution remote-sensing images (HRRSIs), dual-modal HRRSI scene classification is still an active and challenging issue. In this study, we introduce an end-to-end dense-attention–similarity-fusion network (DASFNet) for dual-modal HRRSIs. Specifically, we propose a dense-attention map module based on graph convolution, which adaptively captures long-range semantic cues and further directs shallow-attention cues to the deep level to guide the generation of high-level feature attention cues. At the encoder stage, DASFNet uses feature similarity to explore the correlation between dual-modal features; a similarity-fusion module extracts complementary information by fusing features from different modalities. A multiscale context-feature-aggregation module is used to strengthen the feature embedding of any two spatial scales; this solves the of scale change problem. A large number of experiments on two HRRSI benchmark datasets for scene classification indicate that the proposed DASFNet outperforms the outstanding scene classification approaches.

Critical heat flux (CHF) determines the safety and application of pool boiling heat transfer in many industrial fields. The influence of saturation pressure and surface roughness on saturated pool boiling CHF in hydrofluoroether HFE-7100 was experimentally studied in this investigation. Visualization and heat transfer measurements were conducted at the critical and transition boiling state, and further, the accuracy of CHF prediction models and enhancement mechanism had been analyzed. The polished boiling surfaces had various surface roughness values ranging from 0.019 to 0.587 μm and their contact angles varied from 7° to 10°, while the experimental saturation pressure changed from 0.7 to 2.0 bar. The visual images showed that the pool boiling phenomenon at a critical state was composed of different-sized bubbles, vapor column, and large mushroom vapor, whereas the unsteady blanket of vapor continually injected bubbles at a transition state. The saturation pressure and surface roughness had an obvious improvement on pool boiling CHF, which might be ascribed to the effects of bubble momentum owing to evaporation, distribution and recovery period of a heat transfer boundary layer, capillary action of the working liquid, as well as ratio of vapor jets’ area. Compared with the well-known correlations reported in the literature, CHF correlation of Bailey et al. (2006) predicted the current results more accurately. To further improve the prediction accuracy, a new empirical correlation for CHF dimensionless K considering the effects of saturation pressure and surface roughness was developed, and the predicted values were in better agreement with the experimental data.

Wireless sensor networks (WSNs) are usually used to helps many basic scientific works to gather and observe environmental data, whose completeness and accuracy are the key to ensuring the success of scientific works. However, due to a lot of noise, collision and unreliable data link, data loss and damage in WSNs are rather common. Although some existing works, e.g. interpolation methods or prediction methods, can recover original data to some extent, they maybe provide an unsatisfactory accuracy when the missing data becomes large. To address this problem, this paper proposes a new reliable data transmission scheme in WSNs by using data decomposition and ensemble recovery mechanism. Firstly, the original data are collected by sensor nodes and then are expanded and split into multiple data shares by using multi-ary Vandermonde matrix. Subsequently, these data shares are transmitted respectively to source node via the sensor networks, which is made up of a large number of sensor nodes. Since each share contains data redundancy, the source node can reconstruct the original data even if some data shares are damaged or lost during delivery. Finally, extensive simulation experiments show that the proposed scheme outperforms significantly existing solutions in terms of recovery accuracy and robustness.

Abstract This paper tackles a recent challenge in patrol image processing on how to improve the identification accuracy for power component, especially for the scenarios including many interference objects. Our proposed method can fully use the patrol image information from live work, and it is thus different from traditional power component identification methods. Firstly, we use long short-term memory networks to synthesize the context information in a convolutional neural network. Then, we constructed the Mask LSTM-CNN model by combining the existing Mask R-CNN method and the context information. Further, by extracting the specific features belonging to the power components, we design an optimization algorithm to optimize the parameters of Mask LSTM-CNN model. Our solution is competitive in the sense that the power component is still identified accurately even if the patrol images contain much interference information. Extensive experiments show that the proposed scheme can improve the accuracy of component recognition and has an excellent anti-interference ability. Comparing with the existing R-FCN model and Faster R-CNN model, the proposed method demonstrates a significantly superior detection performance, and the average recognition accuracy is improved from 8 to 11%.

The treatment of diabetes lies in developing novel functional carriers, which are expected to have the unique capability of monitoring blood glucose levels continuously and dispensing insulin correctly and timely. Hence, this study is proposing to create a smart self-regulated insulin delivery system according to changes in glucose concentration. Temperature and glucose dual responsive copolymer microcapsules bearing N-isopropylacrylamide and 3-acrylamidophenylboronic acid as main components were developed by bottom-spray coating technology and template method. The insulinoma β-TC6 cells were trapped in the copolymer microcapsules by use of temperature sensitivity, and then growth, proliferation, and glucose-responsive insulin secretion of microencapsulated cells were successively monitored. The copolymer microcapsules showed favorable structural stability and good biocompatibility against β-TC6 cells. Compared with free cells, the biomicrocapsules presented a more effective and safer glucose-dependent insulin release behavior. The bioactivity of secreted and released insulin did not differ between free and encapsulated β-TC6 cells. The results demonstrated that the copolymer microcapsules had a positive effect on real-time sensing of glucose and precise controlled release of insulin. The intelligent drug delivery system is supposed to mimic insulin secretion in a physiological manner, and further provide new perspectives and technical support for the development of artificial pancreas.

A novel exponential varying-parameter neural network (EVPNN) is presented and investigated to solve the inverse redundancy scheme of the mobile manipulators via quadratic programming (QP). To suspend the phenomenon of drifting free joints and guarantee high convergent precision of the end effector, the EVPNN model is applied to trajectory planning of mobile manipulators. Firstly, the repetitive motion scheme for mobile manipulators is formulated into a QP index. Secondly, the QP index is transformed into a time-varying matrix equation. Finally, the proposed EVPNN method is used to solve the QP index via the matrix equation. Theoretical analysis and simulations illustrate that the EVPNN solver has an exponential convergent speed and strong robustness in mobile manipulator applications. Comparative simulation results demonstrate that the EVPNN possesses a superior convergent rate and accuracy than the traditional ZNN solver in repetitive trajectory planning with a mobile manipulator.

Batch steganography refers to a steganography method where a user tries to hide confidential payload within a batch of images from social networks. It is significantly different from the traditional laboratory steganography where a user only considers an individual image. To apply batch steganography in social media networks, we are faced with two nontrivial problems: 1) how to assign payload to multiple images? and 2) how to recover the hidden payload if some images are lost during transmission? We tackle the problems by: 1) developing an optimal payload embedding strategy and 2) designing a special type of data decomposition. In the former, an optimal non-uniform payload distribution for multiple images is obtained by iterative feature back replacement. In the later, we employ special matrix operation to expand original data and split them into multiple shares. These shares are then embedded into different covers following the optimal non-uniform payload distribution. Our solution is robust in the sense that the recipient can recover the hidden data even if some images are intercepted or lost during delivery. Comprehensive experimental results show that our method outperforms the state-of-the-art in terms of anti-detectability and robustness.

This work tackles a recent challenge in digital image processing: how to identify the steganographic images from a steganographer, who is unknown among multiple innocent actors. The method does not need a large number of samples to train classification model, and thus it is significantly different from the traditional steganalysis. The proposed scheme consists of textural features and clustering ensembles. Local ternary patterns (LTP) are employed to design low-dimensional textural features which are considered to be more sensitive to steganographic changes in texture regions of image. Furthermore, we use the extracted low-dimensional textural features to train a number of hierarchical clustering results, which are integrated as an ensemble based on the majority voting strategy. Finally, the ensemble is used to make optimal decision for suspected image. Extensive experiments show that the proposed scheme is effective and efficient and outperforms the state-of-the-art steganalysis methods with an average gain from 4 % to 6 % .

With the distributed sensors that are deployed to monitor the urban environment in smart cities, the sensed data are overwhelming and beyond the scope of the wireless sensor networks (WSNs) capability. Due to the communication range limits of the sensors, most of these data will be outsourced and stored on some untrusted servers. Thus, how to maintain the data confidentiality and integrity, as well as source authentication and data query privacy of the outsourced data, is a challenging problem. In this paper, we propose a secure searchable encryption scheme, named SSE, for urban sensing and querying to address the problem. Specifically, our SSE constructs a secure hidden vector encryption-(HVE-) based rang query predicate. The sensed data can be stored on an untrusted server in encrypted form. A requester can obtain the correct ciphertexts when his authorized range query matches the HVE-based encryption predicate. With the help of the base station, the ciphertexts can be decrypted and data integrity can be verified; then, the requester can obtain the correct original data. Security analysis demonstrates that; in the SSE, only the authorized requesters can obtain the query results, while the data confidentiality and integrity and source authentication are also preserved.