Search

Mainstream fully supervised deep learning segmentation models can achieve good results when trained on abundant labeled data, but the image segmentation in the medical field faces the challenges of high annotation cost and diverse segmentation targets, often lacking sufficient labeled data. The model proposed in this paper incorporates the idea of extracting labels from data through self-supervision, utilizing superpixels to represent image characteristics for image segmentation under conditions of small sample annotation. The introduction of multiple attention mechanisms allows the model to focus more on spatial features of the image. The position attention module and channel attention module aim to fuse multi-scale features within a single image, while the external attention module highlights the connections between different samples. Experiments were conducted on the CHAOS healthy abdominal organ dataset. In the extreme case of the 1-shot, DSC reached 0.76, which is about 3%higher than the baseline result. In addition, this paper explores the significance of few-shot learning by adjusting the number of N-way-K-shot tasks. Under the 7-shot setting, DSC achieves significant improvement, which is within an acceptable range of the segmentation effect based on full supervision based on deep learning. [ABSTRACT FROM AUTHOR]

Distant supervision relation extraction uses the automatic alignment of natural language texts and knowledge bases to generate labeled training datasets, solving the problem of manual sample labeling. In the current research, most distant supervision does not pay attention to the long-tail data, so most of the sentence bags obtained by distant supervision contain too few sentences. These sentence bags cannot truly and comprehensively express the data itself. This paper proposes a distant supervised relation extraction model (PG+PTATT) based on position-type attention mechanism and graph convolutional network. According to the similarity between sentence bags, Graph Convolutional Networks (GCN) aggregate the implicit high-level features of similar sentence bags to optimize the sentence bags and obtain more prosperous and more comprehensive feature information of the sentence bags. At the same time, an attention mechanism, Position-Type Attention (PTATT) is constructed, which can solve the problem of wrong labels in distant supervision relation extraction: using the position relationships between entity words and non-entity words and type relationships are modeled to reduce the impact of noisy words. The proposed model is experimentally verified on the dataset New York Times (NYT), and the experimental results show that the proposed model can effectively solve the problems existing in distant supervision relation extraction; and it can effectively improve the accuracy of relation extraction.. [ABSTRACT FROM AUTHOR]

With the development of artificial intelligence, digital images have been widely used in various fields. However, due to the appearance of image editing software, a large number of images have been tampered with maliciously, which seriously affects the authenticity of image content. Different from the general object detection, the study of image tampering detection needs to pay more attention to the tamper information of the image itself, which is often manifested in a weak form. Therefore, the detection model needs to focus on learning more abundant tamper features. This paper proposes a dual-stream Faster R-CNN model that combines gradient edge information and attention mechanism, and the model can realize detection and location of regions with different tampering types. One of the two streams is the color stream, which uses the attention mechanism to extract the surface features of the image, such as brightness contrast, visual difference of tampering with the boundary, etc. The second of the two streams is a gradient stream. A Gradient high-pass filter is used to enhance the anomaly edge features between the real area and the tampered area, making it easier for the model to find faint tampered traces in the tampered image. Finally, the features of color stream and gradient stream are fused by means of compact bilinear pooling. Due to the relatively small size of publicly available image tampering data sets, the Pascal VOC 2012 is used to create an image tampering detection data set which containing 10 010 images for model pre-training. The experimental results on COVER, Columbia, and CASIA data sets show that the model proposed in this paper improves the detection accuracy by 7.1% to 9.6% compared to the latest models, and exhibits higher robustness under JPEG compression and image blur attacks. [ABSTRACT FROM AUTHOR]

The maximum mean discrepancy (MMD), as a representative distribution metric between source domain and target domain, has been widely applied in unsupervised domain adaptation (UDA), where both domains follow different distributions, and the labels from source domain are merely available. However, MMD and its class-wise variants possibly ignore the intra-class compactness and interclass separability, thus reducing discriminability of feature representation. This paper proposes a focally discriminative loss for unsupervised domain adaptation. This method endeavors to improve the discriminative ability of MMD from two aspects: (1) the weights are re-designed for MMD in order to align the distribution of relatively hard classes across domains; (2) a focally contrastive loss is explored to tradeoff the positive sample pairs and negative ones for better discrimination. The integration of both losses can not only make the intra-class features close, but also push away the inter-class features far from each other. Moreover, the improved loss is simple yet effective, and it can be extended to the network structure of the attention mechanism. Experiments on several domain adaptation datasets verify the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

In the field of music information retrieval (MIR), classification according to music genres is a challenging task. Traditional audio feature engineering methods requires manually selecting and extracting music signal features for processing, resulting in complex feature extraction process, unstable model performance and poor generalization. The method combining deep learning with spectrogram also has some problems such as unsuitable model for some data and difficulty in global feature extraction. This paper proposes a music genre classification model based on convolutional attention mechanism, called MGTN. MGTN combines two music genre classification methods: input spectrogram and audio signal feature extraction, to construct audio time series data, which greatly improves the model's ability to extract features and generalization, and provides a new idea for music genre classification. Experimental results on GTZAN and Ballroom datasets show that the MGTN model can effectively fuse input data from two different modalities. Compared with dozens of benchmark models, the MGTN model has strong advantages. [ABSTRACT FROM AUTHOR]

In the driving scenario, due to the occlusion between pedestrians and their scale variations, detection model have low accuracy, high model parameters, and difficulty in deploying to mobile terminals. This paper proposes a lightweight real-time pedestrian detection model, LPD-YOLO, based on the YOLOv5s model. Firstly, in the feature extraction part, the original backbone network is replaced with MES Net (Mish-Enhanced Shuffle Net), and an attention module SA (Shuffle Attention) is embedded in the backbone network to enhance network feature extraction ability. Secondly, in the feature fusion part, the original PANet is improved by using the DS-ASFF structure to fully fuse feature maps of different sizes. Then, standard convolution is replaced with GS convolution in the feature-covergent network part without affecting accuracy, further reducing model parameters and computation. Finally, in the prediction part, the original loss function is improved by using the OTA label assignment strategy combined with α-IOU to accelerate model convergence. Experimental data show that compared with YOLOv5s, LPD-YOLO has 81.2% fewer parameters, 46.3% lower floating-point operation volume, 75.8% smaller model size, and 3.3% higher detection accuracy. The single image detection speed is 13.2 ms, which better meets the real-time detection requirements of dense pedestrians in driving scenarios. [ABSTRACT FROM AUTHOR]

Aspect-level sentiment classification aims to discriminate the sentiment polarity of a specific aspect in a sentence. Although attention-based recurrent neural network models perform well among existing solutions, they are not ideal for processing "semantically ambiguous" sentences that are short and contain many neologisms and polysemous words. Therefore, this paper proposes a neural network model based on knowledge graph and attention mechanism. The basic idea is to use a knowledge base to obtain a relevant concept set of aspect words and integrate external information to enhance the semantic representation of the text. Firstly, the output of bidirectional long short-term memory network is combined with self-attention mechanism to generate context representation. Then, the upper and lower context representations are combined to use dual attention to obtain external knowledge from the knowledge graph and obtain knowledge vectors related to aspect words. Finally, the two parts of content are input together into a fully connected network to calculate the aspect-level sentiment tendency. Experimental results show that compared with other models, the proposed model significantly improves classification performance. [ABSTRACT FROM AUTHOR]

Aiming at the problem of large number of parameters and low detection efficiency of the traditional target detection method in defect detection of electronic components, this paper proposes a target detection method based on the lightweight YOLOX detection network. Firstly, the backbone network is lightened using deeply separable convolution to reduce parameters and improve detection speed. Secondly, a spatial Pyramid-based channel attention model is constructed to filter and fuse features of different scales to enhance the feature weights of small size defects. In the feature fusion upsampling process, efficient channel attention is added to improve detection accuracy with slightly increased parameters. Finally, the EIoU loss function is used to optimize the IoU loss function, and the cosine annealing algorithm is used to make the model achieve the best detection effect. The model is tested on a self-made dataset of electronic component appearance defects, and the average detection accuracy reaches 98.96%, with a detection time of approximately 0.09 seconds per image. Compared with the original model, the detection speed is doubled and the model size is reduced by about 60%. The model is also validated on the PCB defect public dataset, achieving fast detection of target defects. [ABSTRACT FROM AUTHOR]

Air pollution seriously affects human health and social sustainable development. However, the multi-variable air quality data obtained by sensors often have missing values, which brings difficulties to data analysis and processing. Currently, many analysis methods for changes in a certain air component only rely on time data and spatial data of this attribute, ignoring the influence of other air components on the trend of this attribute in the same time interval. In addition, it is difficult to achieve ideal results in filling discrete missing data. This paper proposes a Time Attention Model (TAM) based on deep learning, which uses attention mechanism to focus on the correlation between different timestamps and the correlation between different feature time series, and combines short-term historical data to fill missing values in multi-variable air quality data. The proposed model is evaluated using air quality data from Beijing, and the experimental results show that TAM has advantages over ten other baseline models. [ABSTRACT FROM AUTHOR]

Attention mechanism has recently shown superior performance in deep neural networks, its computation generates complex data flow and requires high computation and memory overheads. Therefore, customized accelerators are required to optimize the inference computing. This paper proposes an accelerator architecture for attention mechanism computation. A flexible partitioning method based on hardware control is used to divide the huge matrices in the attention model into hardwarefriendly computing blocks, which realizes the systolic array in accelerator matched by the block computation match. A layer fusion computing structure based on two-step softmax function decomposition is proposed, which effectively reduces the memory access of attention mechanism computation. A fusedlayer attention model accelerator based on fine-grained computational scheduling is designed and implemented by HDL. The performance was evaluated based on the XLINIX FPGA device and HLS tool. Compared with the CPU and GPU implementation under the same settings, the delay of accelerator was improved by 4.91 times, the efficiency of accelerator was improved by 1.24 times. [ABSTRACT FROM AUTHOR]

Knowledge Graph (KG) owns rich structured information, which can effectively alleviate the sparsity and cold start problem of recommendation model and improve the accuracy and interpretability of recommendation. In recent years, the end-to-end recommendation models incorporating knowledge graph have become a technological trend. This paper proposes a KG recommendation model incorporating the influence effects of similar users. The proposed method expands the interaction between users and items to effectively utilize the knowledge graph. Firstly, a graph neural network neighbor aggregation strategy and an attention mechanism are used to capture two higher-order representations of users and items on the knowledge graph respectively. Then, an influence enhancement layer is designed to capture potential representations of similar users􀆳 influence effects according to their influence effects. Finally, these three representations are fed together into a multi-layer perceptron to output prediction scores. Experimental results on real datasets show the effectiveness and efficiency of the proposed model. [ABSTRACT FROM AUTHOR]

Aiming at the problems of edge confusion caused by multiple objects gathering in remote sensing image, unclear segmentation of small scale objects, and insufficient global information acquisition in semantic segmentation process, this paper proposes a semantic segmentation algorithm of remote sensing images based on mixed attention and full-scale skip connection network, called DU-net. In this algorithm, U-net3+ is used as the basic network, and full-scale skip connection network is used as the feature extraction network. The depth supervision in the original model is abandoned, the association between feature and attention mechanism is established, and the process of semantic segmentation is finally realized. The experimental results show that the DU-net algorithm has significant improvement over the classical algorithm under different indexes, and improves the quality of image edge segmentation and the accuracy of the algorithm for small scale target segmentation. [ABSTRACT FROM AUTHOR]

As one of the research hotspots of object detection, anchor free abandons a large number of predefined box Settings and adopts pixel-by-pixel method for prediction. Even so, it does not deal well with overlapping objects. In addition, the ability of network to obtain global information of images is weak and receptive field mismatch is easy to occur. Therefore, this paper proposes two modules: diverse receptive field attention mechanism (DRAM) and global context-guided feature fusion module (GCF). Extensive experiments on the PASCAL VOC and MS COCO confirm the effectiveness of our method. Compared with the baseline FCOS, the proposed methodcanimprovePASCAL VOCby1.4 pointsandobtaina mAPof42.8on MSCOCO. The detection performanceis significantly beter than many advanced algorithms. [ABSTRACT FROM AUTHOR]

Personalized recommendation systems are widely used in reducing information overload, providing personalized services, and assisting users in decision-making. Link prediction is one of the important methods of personalized recommendation. Traditional heuristic link prediction methods only consider the graph structure characteristics of the network, and lack the application of explicit and implicit feature information, and most methods are based on undirected and unweighted networks. Aiming at the shortcomings of traditional link prediction methods, this paper proposes a link prediction method AFP based on the collective attention flow network and the R-GCN method. The different edge directions between the two nodes in the attention flow network are abstracted into two types of edge relations. The attention mechanism is introduced to learn the node attributes and edge attributes in the network, and the network's graph structure characteristics, implicit characteristics and explicit features are comprehensively considered. The scoring function is used to get the probability of the establishment of the triple, and the link prediction problem is transformed into a two-category problem, thus predicting the possibility that the edges between nodes belong to a certain type of relationship. Experiments show that, compared with 6 benchmark models such as GCN and GAT, this method improves the accuracy, precision, recall and other evaluation indicators. [ABSTRACT FROM AUTHOR]

Based on Generative Adversarial Networks (GANs), most current text-to-image generation algorithms focus on designing different attention generation models to improve the characterization and expression of image details. However, they ignore the discriminator􀆳s perception of key local semantics, so the generation models can easily generate poor image details to "fool" the discriminators. This paper designs a vocabulary-image discriminative attention module in the discriminators to enhance the discriminators􀆳 ability to perceive and capture key semantics, and drive the generation model to generate high-quality image details. Therefore, a discrimination-enhanced generative adversarial model (DE-GAN) is proposed. The experimental results show that, on the CUB-Bird dataset, DE-GAN achieves 4.70 on the IS index, which is 4.2% higher than the baseline model and achieves high performance. [ABSTRACT FROM AUTHOR]

Identifying potential academic rising stars from academic newcomers can provide decision support for tasks such as talent introduction, project review, and expert database construction, which has important research significance and application value and has received extensive attention from the academic community. However, the existing academic rising star prediction methods do not organically combine the academic cooperation relationship and individual attribute information, resulting in low accuracy. To solve the above problem, this paper proposes an academic rising star prediction method MGCNA based on multi-graph convolutional neural network and attention mechanism. It comprehensively considers cooperative networks and similar networks. Based on the two networks, the graph convolutional neural network is used to learn the author􀆳s feature representation, and then the attention mechanism is used for information fusion, so as to predict the academic rising stars with high potential. Finally, experiments are carried out on real datasets from the ArnetMiner platform, and the experimental results demonstrate the effectiveness of MGCNA in predicting academic rsing star tasks. [ABSTRACT FROM AUTHOR]

The standardization of Chinese geographic addresses plays a crucial role in the current construction of smart cities. The traditional geographic address standardization technology usually uses the methods of similarity calculation or rule base matching based on the text character level, and the processing effect of complex, special or redundant addresses is poor. This paper proposes an address matching algorithm that combines attention mechanism and multi-level representation by converting the address standardization task into a matching degree calculation task for similar addresses. Firstly, according to the special grammatical structure of the address text, a standard address tree is constructed by using the Trie grammatical tree. Secondly, based on the attention mechanism, the Bi-LSTM network and the CNN network are used to generate multi-level semantic representations of address pairs. Finally, the similarity is calculated by Manhattan distance. On the self-built dataset, the proposed SGAM (Symmetrical Geographic Address Matching) model improves the matching accuracy (91.22%) by 4%~10% in comparison to TextRCNN, FastText, attention-based convolutional neural network (ABCNN) and other models, proving that the SGAM model has better performance on the address matching task. [ABSTRACT FROM AUTHOR]

Semantic segmentation algorithms can classify images a t the pixel level, and are widely used in fields such as unmanned driving, medical image processing, and industrial automation, and h ave important research value. The research of semantic segmentation algorithms focuses on three aspects: improving the accuracy of segmentation, reducing the amount of parameter s and increasing the speed of inference. The lightweight semantic segmentation algorithm ENet uses a multi-layer convolutional codec and a large number of dilated convolutions to avoid excessive downsampling and use of spatial information. Although it retains some spatial information integrity and large receptive field, the codec is bloated, the transmission of spatial information is poor, and the sensory field overflows and causes gr id effect. Aiming at the above problems, this paper tailors the EN et algorithm structure, uses the attention mechanism and the pyramid dilated convolution to design spatial information transmission module, optimizes the algorithm structure, improve s the algorithm receptive field, and completely transmits the spatial information transmission. The experimental results on public datasets Cityscapes and BDD100K show that the new module can improve the performance of the original algorithm with a smaller amount of parameters and calculations, which proves the redundancy of t he original algorithm and the effectiveness of the designed module. [ABSTRACT FROM AUTHOR]