Graphs have been widely adopted to accomplish fraud detection tasks because of their inherently favorable structure capture the intricate features in many complicated scenarios, especially some modern e-commerce situations that various relation attributes like transactions. These works tend utilize direct aggregate information about neighbor nodes target node or aggregation after conditional filter and mostly use local but ignore global information. However, cases, abnormal points are...

Purpose The purpose of this study is to solve the problems poor stability and high energy consumption dynamic window algorithm (DWA) for mobile robots, a novel enhanced proposed in paper. Design/methodology/approach takes distance function as weight target-oriented coefficient, new evaluation presented optimize azimuth angle. Findings jitter robot caused by drastic change angular velocity reduced when closer target point. simulation results show that effectively optimizes during operation...

Concept drift is a phenomenon where the distribution of data streams changes over time. When this happens, model predictions become less accurate. Hence, models built in past need to be re-learned for current data. Two design questions addressed designing strategy re-learn models: which type concept has occurred, and how utilize improve re-learning performance. Existing detection methods are often good at determining when occurred. However, few retrieve information about came present stream....

In recent years, graph-based fraud detection methods have garnered increasing attention for their superior ability to tackle the issue of camouflage in fraudulent scenarios. However, these often rely on a substantial proportion samples as training set, disregarding reality scarce annotated real-life As theoretical framework within semi-supervised learning, principle consistency regularization posits that unlabeled should be classified into same category own perturbations. Inspired by this...

As a type of evolving-fuzzy system, the evolving-fuzzy-neuro (EFN) system uses structure inspired by neural networks to determine its parameters (fuzzy sets and fuzzy rules), so EFN can inherit advantages networks. However, for streaming data regression, systems still have several drawbacks: determining is not robust sequence; rules complex as subspaces that approximate Takagi–Sugeno–Kang (TSK) rule need be obtained, many optimized; it difficult detect adapt changes in distribution, i.e.,...

Since support vector regression (SVR) is a flexible algorithm, its computational complexity does not depend on the dimensionality of input space, and it has excellent generalization capability. However, central assumption with SVRs that all required data available at time construction, which means these algorithms cannot be used streams. Incremental SVR been offered as potential solution, but accuracy suffers noise learning speeds are slow. To overcome two limitations, we propose novel...

In order to extract useful information from data streams, incremental learning has been introduced in more and mining algorithms. For instance, a self-organizing neural network (SOINN) proposed topological structure that consists of one or networks closely reflect the distribution streams. However, SOINN tradeoff between deleting previously learned nodes inserting new nodes, i.e., stability-plasticity dilemma. Therefore, it is not guaranteed obtained by will represent distribution. solving...

When a train leaves platform, knowing the carriage load (the number of passengers in each carriage) this will support managers to guide at next platform choose carriages avoid congestion. This capacity has become critical since onset pandemic. However, with dynamicity and speed trains improved (about 3 minutes travel between stations) as well station stop period reduced (60–90 second per station), real-time prediction is more challenging. paper presents an intelligent system, which developed...

Detection of planar surfaces in a generic scene is difficult when the illumination complex and less intense, have non-uniform colors (e.g., movie poster). As result, specularity, if appears, superimposed with surface color pattern, hence observation uniform specularity no longer sufficient for identifying as it does under distant point light source. In this paper, we address problem recognition single generic-scene image. particular, study recaptured photo an application image forensics. We...