Graph-based fraud detection approaches have escalated lots of attention recently due to the abundant relational information of graph-structured data, which may be beneficial for the detection of fraudsters. However, the GNN-based algorithms could fare poorly when the label distribution of nodes is heavily skewed, and it is common in sensitive areas such as financial fraud, etc. To remedy the class imbalance problem of graph-based fraud detection, we propose a Pick and Choose Graph Neural Network (PC-GNN for short) for imbalanced supervised learning on graphs. First, nodes and edges are picked with a devised label-balanced sampler to construct sub-graphs for mini-batch training. Next, for each node in the sub-graph, the neighbor candidates are chosen by a proposed neighborhood sampler. Finally, information from the selected neighbors and different relations are aggregated to obtain the final representation of a target node. Experiments on both benchmark and real-world graph-based fraud detection tasks demonstrate that PC-GNN apparently outperforms state-of-the-art baselines. CCS CONCEPTS• Computing methodologies → Neural networks; • Security and privacy → Software and application security.
The rapid proliferation of new users and items on the social web has aggravated the gray-sheep user/long-tail item challenge in recommender systems. Historically, cross-domain co-clustering methods have successfully leveraged shared users and items across dense and sparse domains to improve inference quality. However, they rely on shared rating data and cannot scale to multiple sparse target domains (i.e., the one-to-many transfer setting). This, combined with the increasing adoption of neural recommender architectures, motivates us to develop scalable neural layer-transfer approaches for cross-domain learning. Our key intuition is to guide neural collaborative filtering with domain-invariant components shared across the dense and sparse domains, improving the user and item representations learned in the sparse domains. We leverage contextual invariances across domains to develop these shared modules, and demonstrate that with user-item interaction context, we can learn-to-learn informative representation spaces even with sparse interaction data. We show the effectiveness and scalability of our approach on two public datasets and a massive transaction dataset from Visa, a global payments technology company (19% Item Recall, 3x faster vs. training separate models for each domain). Our approach is applicable to both implicit and explicit feedback settings.
At present, intracranial pressure (ICP) can be reliably measured directly by invasive techniques such as ventriculostomy tube 1,2 , which require special equipment for the placement of monitoring probes. The most common side effect is intracranial infection. 3-5 lumbar puncture (lP) is one of the most common methods of ICP measurement, however, when ICP is very high, it should be performed with caution to avoid inducing brain herniation which may lead to death. 6 In recent years, some noninvasive ICP monitoring methods have been reported, including transcranial doppler, 7-9 jugular venous oxygen concentration monitoring, measurement of tympanic membrane displacement, 10,11 and retinal venous pressure measurement, however, none of these methods is sufficiently accurate to be allowed for routine clinical use.As we know, it is now possible to record and monitor the continuous digital EEG, and continuous EEG monitoring provides dynamic information about brain function. 12 Electroencephalogram (EEG) power spectrum is a method of EEG signal analysis which provides more accurate information for determining brain functions than traditional EEG techniques. 13,14 Electroencephalogram power spectrum has been used to assess the level of consciousness and the depth of anesthesia. 15 Gradual increase of ICP will lead to change of ABSTRACT: Objectives: To investigate the feasibility of Electroencephalogram (EEG) power spectrum analysis as a noninvasive method for monitoring intracranial pressure (ICP). Methods: The EEG signals were recorded in 62 patients (70 cases) with central nervous system (CNS) disorders in our hospital. By using self-designed software, EEG power spectrum analysis was conducted and pressure index (PI) was calculated automatically. Intracranial pressure was measured by lumbar puncture (lP). Results: We found a significant negative correlation between PI and ICP (r = -0.849, p < 0.01). Conclusions: The PI obtained from EEG analysis is correlated with ICP. Analysis of specific parameters from EEG power spectrum might reflect the ICP.RÉSUMÉ: Une nouvelle méthode de surveillance de la pression intracrânienne par analyse des spectres de puissance de l'EEG. Objectifs : le but de l'étude était d'évaluer la faisabilité de l'analyse des spectres de puissance de l'EEG comme méthode non effractive de surveillance de la pression intracrânienne (PIC). Méthode : les signaux EEG ont été enregistrés chez 62 patients (70 observations) atteints de troubles du système nerveux central (SNC) dans notre hôpital. Nous avons effectué une analyse des spectres de puissance de l'EEG et l'indice de pression (IP) a été calculé automatiquement. la pression intracrânienne a été mesurée par ponction lombaire. Résultats : Nous avons observé une corrélation négative significative entre l'IP et la PIC (r = -0,849 ; p < 0,01).
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.