Graph Convolutional Neural Networks (graph CNNs) have been widely used for graph data representation and semi-supervised learning tasks. However, existing CNNs generally use a fixed which may not be optimal In this paper, we propose novel Learning-Convolutional Network (GLCN) learning. The aim of GLCN is to learn an structure that best serves by integrating both convolution in unified network architecture. main advantage given labels the estimated are incorporated thus can provide useful...

In recent years, more and people are paying close attention to the environmental problems in metropolitan areas their harm human body. Among them, haze is pollutant that most concerned about. The demand for a method predict level public academics keeps rising. order concentration on time scale hours, this study built prediction based one-dimensional convolutional neural networks. gated recurrent unit was used comparison, which highlights training speed of network. summary, data past 24 h as...

Previous studies on deep learning (DL) applications in pathology have focused pathologist-versus-algorithm comparisons. However, DL will not replace the breadth and contextual knowledge of pathologists; rather, only through their combination may benefits be achieved. A fully crossed multireader multicase study was conducted to evaluate assistance with pathologists' diagnosis gastric cancer. total 110 whole-slide images (WSI) (50 malignant 60 benign) were interpreted by 16 board-certified...

Graph Convolutional Networks (GCNs) have been widely studied for semi-supervised learning tasks. It is known that the graph convolution operations in most of existing GCNs are composed two parts, i.e., feature propagation (FP) on a neighborhood and transformation (FT) with fully connected network. For learning, generally utilize label information only to train parameters FT part via optimizing loss function. However, they lack exploiting propagation. Besides, due fixed topology used FP,...

At present, feature-based 3D reconstruction and tracking technology is widely applied in the medical field. In minimally invasive surgery, surgeon can achieve three-dimensional through images obtained by endoscope human body, restore scene of area to be operated on, track motion soft tissue surface. This enables doctors have a clearer understanding location depth surgical area, greatly reducing negative impact 2D image defects ensuring smooth operation. this study, firstly, coordinates each...

Graph Neural Networks (GNNs) are gaining increasing attention on graph data learning tasks in recent years. However, many applications, may be coming an incomplete form where attributes of nodes partially unknown/missing. Existing GNNs generally designed complete graphs which can not deal with attribute-incomplete directly. To address this problem, we develop a novel partial aggregation based GNNs, named Partial (PaGNNs), for representation and learning. Our work is motivated by the...

Graph Attention (GA) which aims to learn the attention coefficients for graph edges has achieved impressive performance in GNNs on many learning tasks. However, existing GAs are usually learned based edges' (or connected nodes') features fail fully capture rich structural information of edges. Some recent research attempts incorporate into GA but how exploit them is still a challenging problem. To address this challenge, work, we propose leverage new Replicator Dynamics model learning,...

Recently, deep graph matching (GM) methods have gained increasing attention. These integrate nodes¡¯s embedding, node/edges¡¯s affinity learning and final correspondence solver together in an end-to-end manner. For problem, one main issue is how to generate consensus node's embeddings for both source target graphs that best serve tasks. In addition, it also challenging incorporate the discrete one-to-one constraints into differentiable network. To address these issues, we propose a novel...

Previous studies have shown the effectiveness of deep learning algorithms in improving detection credit card fraud, which has become a major issue for financial institutions. This paper discusses how two methods, Convolutional Neural Network (CNN) and auto-encoder, perform fraud task using different datasets. research utilizes brand-new dataset with all raw input variables Université Libre de Bruxelles (ULB) transaction dataset, been preprocessed PCA technology. Since imbalanced datasets can...

Recently, graph Convolutional Neural Networks (graph CNNs) have been widely used for data representation and semi-supervised learning tasks. However, existing CNNs generally use a fixed which may be not optimal In this paper, we propose novel Graph Learning-Convolutional Network (GLCN) learning. The aim of GLCN is to learn an structure that best serves by integrating both convolution together in unified network architecture. main advantage GLCN, given labels the estimated are incorporated...

Graph contrastive learning is usually performed by first conducting Data Augmentation (GDA) and then employing a pipeline to train GNNs. As we know that GDA an important issue for graph learning. Various GDAs have been developed recently which mainly involve dropping or perturbing edges, nodes, node attributes edge attributes. However, our knowledge, it still lacks universal effective augmentor suitable different types of data. To address this issue, in paper, introduce the message...

Graph Convolutional Networks (GCNs) gain increasing attention on graph data learning tasks in recent years. However, many applications, may come with an incomplete form where attributes of nodes are partially unknown/missing. Existing Convolutions (GCs) generally designed complete graphs which can not deal attribute-incomplete directly. To address this problem, paper, we extend standard GC and develop explicit Partial Convolution (PaGC) for data. Our PaGC is derived based the observation...

In recent years, graph prompt learning/tuning has garnered increasing attention in adapting pre-trained models for representation learning. As a kind of universal learning method, Graph Prompt Feature (GPF) achieved remarkable success Neural Networks (GNNs). By fixing the parameters GNN model, aim GPF is to modify input data by adding some (learnable) vectors into node features better align with downstream tasks on smaller dataset. However, existing GPFs generally suffer from two main...

Graph Convolutional Networks (GCNs) have been widely studied. The core of GCNs is the definition convolution operators on graphs. However, existing Convolution (GC) are mainly defined adjacency matrix and node features generally focus obtaining effective embeddings which cannot be utilized to address graphs with (high-dimensional) edge features. To this problem, by leveraging tensor contraction representation product graph diffusion theories, paper analogously defines an operator named as...

Abstract In this article, the integrity of seismic catalog obtained (1970–2014, M > 2.8) was verified according to Gutenberg–Richter relation, appropriate minimum magnitude determined, and whole region divided into five areas geological structure background research object trend historical zone. We applied multifractal analysis in each partition. The results showed that although different backgrounds, before major earthquakes, earthquake time series information dimension had degrees...

Modern marketing plays a vital role in promoting the development of enterprises.Enterprises need to plan and strategically carry out activities based on products, channels, promotion price.In entire process, packaging products is an important part.Packaging modern society not just for protecting facilitating product storage.Exquisite personalized one crucial component expanding sales.This article will mainly discuss relationship between design.

Graph Convolutional Networks (GCNs) have been widely demonstrated their powerful ability in graph data representation and learning. Existing convolution layers are mainly designed based on signal processing transform aspect which usually suffer from some limitations, such as over-smoothing, over-squashing non-robustness, etc. As we all know that Convolution Neural (CNNs) received great success many computer vision machine One main is CNNs leverage learnable filters (kernels) to obtain rich...

Graph Convolutional Representation (GCR) has achieved impressive performance for graph data representation. However, existing GCR is generally defined on the input fixed which may restrict representation capacity and also be vulnerable to structural attacks noises. To address this issue, we propose a novel Latent (LatGCR) robust learning. Our LatGCR derived based reformulating convolutional from aspect of neighborhood reconstruction. Given an $\textbf{A}$, aims generate flexible latent...

Abstract The Internet era has brought great changes to people’s lives, it promotes the continuous development of economy, but also promote overall upgrade brand, so that brand visual image communication more media. Under technology development, requirement design is higher than before, can not only limited level, at same time combine with modern new way media based on culture, and enhance experience. Brand be feasible, promoting implementation. This paper will mainly discuss path under...

Graph Convolutional Networks (GCNs) have been commonly studied for graph learning tasks, such as semi-supervised learning, clustering etc. However, many existing GCNs are generally conducted on single data and thus can not be applied directly to multi-graph that consists of various types edges between nodes. To address this issue, in paper, we propose a novel multiple Adversarial Regularized Learning (mGARL) framework representation learning. mGARL aims learn an optimal structure...

To alleviate the local receptive issue of GCN, Transformers have been exploited to capture long range dependences nodes for graph data representation and learning. However, existing generally employ regular self-attention module all node-to-node message passing which needs learn affinities/relationships between node's pairs, leading high computational cost issue. Also, they are usually sensitive noises. overcome this issue, we propose a novel Transformer architecture, termed Anchor Graph...