The problem of session-based recommendation aims to predict user actions based on anonymous sessions. Previous methods model a session as sequence and estimate representations besides item make recommendations. Though achieved promising results, they are insufficient obtain accurate vectors in sessions neglect complex transitions items. To embedding take items into account, we propose novel method, i.e. Session-based Recommendation with Graph Neural Networks, SR-GNN for brevity. In the...

Graph representation learning nowadays becomes fundamental in analyzing graph-structured data. Inspired by recent success of contrastive methods, this paper, we propose a novel framework for unsupervised graph leveraging objective at the node level. Specifically, generate two views corruption and learn representations maximizing agreement these views. To provide diverse contexts objective, hybrid scheme generating on both structure attribute levels. Besides, theoretical justification behind...

Session-based recommendation nowadays plays a vital role in many websites, which aims to predict users' actions based on anonymous sessions. There have emerged studies that model session as sequence or graph via investigating temporal transitions of items session. However, these methods compress into one fixed representation vector without considering the target be predicted. The will restrict ability recommender model, diversity and interests. In this paper, we propose novel attentive...

Recently, contrastive learning (CL) has emerged as a successful method for unsupervised graph representation learning. Most CL methods first perform stochastic augmentation on the input to obtain two views and maximize agreement of representations in views. Despite prosperous development methods, design schemes -- crucial component remains rarely explored. We argue that data should preserve intrinsic structures attributes graphs, which will force model learn are insensitive perturbation...

Mapping the connectome of human brain using structural or functional connectivity has become one most pervasive paradigms for neuroimaging analysis. Recently, Graph Neural Networks (GNNs) motivated from geometric deep learning have attracted broad interest due to their established power modeling complex networked data. Despite superior performance in many fields, there not yet been a systematic study how design effective GNNs network To bridge this gap, we present BrainGB, benchmark analysis...

Vascular inflammation is well known for its ability to compromise the function of blood--brain barrier (BBB). Whether on parenchymal side barrier, such as that associated with Parkinson's-like dopamine (DA) neuron lesions, similarly disrupts BBB function, unknown. We assessed integrity by examining leakage FITC-labeled albumin or horseradish peroxidase from vasculature into parenchyma in animals exposed DA neurotoxin 6-hydroxydopamine (6OHDA). Unilateral injections 6OHDA striatum medial...

Graph convolutional networks (GCNs) have been successfully applied in node classification tasks of network mining. However, most these models based on neighborhood aggregation are usually shallow and lack the “graph pooling” mechanism, which prevents model from obtaining adequate global information. In order to increase receptive field, we propose a novel deep Hierarchical Convolutional Network (H-GCN) for semi-supervised classification. H-GCN first repeatedly aggregates structurally similar...

Multimedia contents are of predominance in the modern Web era. Recent years have witnessed growing research interests multimedia recommendation, which aims to predict whether a user will interact with an item multimodal contents. Most previous studies focus on modeling user-item interactions features included as side information. However, this scheme is not well-designed for recommendation. First, only <italic xmlns:mml="http://www.w3.org/1998/Math/MathML"...

Deep learning has achieved remarkable success in representations for molecules, which is crucial various biochemical applications, ranging from property prediction to drug design. However, training Neural Networks (DNNs) scratch often requires abundant labeled are expensive acquire the real world. To alleviate this issue, tremendous efforts have been devoted Chemical Pre-trained Models (CPMs), where DNNs pre-trained using large-scale unlabeled molecular databases and then fine-tuned over...

Graphs are widely used to describe real-world objects and their interactions. Graph Neural Networks (GNNs) as a de facto model for analyzing graphstructured data, highly sensitive the quality of given graph structures. Therefore, noisy or incomplete graphs often lead unsatisfactory representations prevent us from fully understanding mechanism underlying system. In pursuit an optimal structure downstream tasks, recent studies have sparked effort around central theme Structure Learning (GSL),...

Graph Contrastive Learning (GCL) establishes a new paradigm for learning graph representations without human annotations. Although remarkable progress has been witnessed recently, the success behind GCL is still left somewhat mysterious. In this work, we first identify several critical design considerations within general paradigm, including augmentation functions, contrasting modes, contrastive objectives, and negative mining techniques. Then, to understand interplay of different...

Internet of Things (IoT) is becoming truly ubiquitous in our everyday life, but it also faces unique security challenges. Intrusion detection critical for the and safety a wireless IoT network. This paper discusses human-in-the-loop active learning approach intrusion detection. We first present fundamental challenges against design successful Detection System (IDS) then briefly review rudimentary concepts propose its employment diverse applications Experimental example presented to show...

Mitochondria are the source of reactive oxygen species (ROS) in plant cells and play a central role mitochondrial electron transport chain (ETC) tricarboxylic acid cycle (TCA) cycles; however, ROS production regulation for seed germination, seedling growth, as well responses to abiotic stress, not clear. This study was conducted obtain basic information on embryo antioxidant responses, protein profile changes artificial aging oat seeds (Avena sativa L.) exposed exogenous nitric oxide (NO)...

Acclimation of plants to adverse conditions requires the coordination gene expression and signalling pathways between tissues cell types. As energy carbon capturing organs, leaves are significantly affected by abiotic biotic stresses. However, tissue- or type-specific analyses stress responses have focussed on Arabidopsis root. Here, we comparatively explore transcriptomes three leaf (epidermis, mesophyll, vasculature) after induction diverse chemical stimuli (antimycin A,...

Pretrained Language Models (PLMs) such as BERT have revolutionized the landscape of Natural Processing (NLP). Inspired by their proliferation, tremendous efforts been devoted to Graph (PGMs). Owing powerful model architectures PGMs, abundant knowledge from massive labeled and unlabeled graph data can be captured. The implicitly encoded in parameters benefit various downstream tasks help alleviate several fundamental issues learning on graphs. In this paper, we provide first comprehensive...

Multimodal brain networks characterize complex connectivities among different regions from both structural and functional aspects provide a new means for mental disease analysis. Recently, Graph Neural Networks (GNNs) have become de facto model analyzing graph-structured data. However, how to employ GNNs extract effective representations in multiple modalities remains rarely explored. Moreover, as no initial node features, design informative attributes leverage edge weights learn is left...

Next point-of-interest (POI) recommendation aims to predict a user's next destination based on sequential check-in history and set of POI candidates. Graph neural networks (GNNs) have demonstrated remarkable capability in this endeavor by exploiting the extensive global collaborative signals present among POIs. However, most existing graph-based approaches construct graph structures pre-defined heuristics, failing consider inherent hierarchical features such as geographical locations...

Training deep neural networks (DNNs) with limited supervision has been a popular research topic as it can significantly alleviate the annotation burden. Self-training successfully applied in semi-supervised learning tasks, but one drawback of self-training is that vulnerable to label noise from incorrect pseudo labels. Inspired by fact samples similar labels tend share representations, we develop neighborhood-based sample selection approach tackle issue noisy We further stabilize via...

Seeds lose their viability when they are exposed to high temperature and moisture content (MC) during storage. The expression metabolism of proteins plays a critical role in seed resistance heat stress. However, the proteome response stress oat (Avena sativa) seeds storage has not been revealed. To understand mechanisms acclimation tolerance seeds, an integrated physiological comparative proteomic analysis was performed on with different MC Oat 10% 16% were subjected temperatures (35, 45,...

Brain networks characterize complex connectivities among brain regions as graph structures, which provide a powerful means to study connectomes. In recent years, neural have emerged prevalent paradigm of learning with structured data. However, most network datasets are limited in sample sizes due the relatively high cost data acquisition, hinders deep models from sufficient training. Inspired by meta-learning that learns new concepts fast training examples, this paper studies data-efficient...