Search

Abstract The variability of sea surface temperatures (SSTs) is crucial in climate dynamics, influencing marine ecosystems and human activities. This study leverages graph neural networks (GNNs), specifically a GraphSAGE model, to forecast SSTs and their anomalies (SSTAs), focusing on the global scale structure of climatological data. We introduce an improved graph construction technique for SST teleconnection representation. Our results highlight the GraphSAGE model's capability in 1‐month‐ahead global SST and SSTA forecasting, with SST predictions spanning up to 2 years with a recursive approach. Notably, regions with persistent currents exhibited enhanced SSTA predictability, contrasting with equatorial and Antarctic areas. Our GNN outperformed both the persistence model and traditional methods. Additionally, we offer an SST and SSTA graph data set based on ERA5 and a graph generation tool. This GNN case study has shown how the GraphSAGE can be used in SST and SSTA forecasting, and will provide a foundation for further refinements such as adjusting graph construction, optimizing imbalanced regression techniques for extreme SSTAs, and integrating GNNs with other temporal pattern learning methods to improve long‐term predictions.

By developing awareness of smartphone activities that the user is performing on their smartphone, such as scrolling feeds, typing and watching videos, we can develop application features that are beneficial to the users, such as personalization. It is currently not possible to access real-time smartphone activities directly, due to standard smartphone privileges and if internal movement sensors can detect them, there may be implications for access policies. Our research seeks to understand whether the sensor data from existing smartphone inertial measurement unit (IMU) sensors (triaxial accelerometers, gyroscopes and magnetometers) can be used to classify typical human smartphone activities. We designed and conducted a study with human participants which uses an Android app to collect motion data during scrolling, typing and watching videos, while walking or seated and the baseline of smartphone non-use, while sitting and walking. We then trained a machine learning (ML) model to perform real-time activity recognition of those eight states. We investigated various algorithms and parameters for the best accuracy. Our optimal solution achieved an accuracy of 78.6% with the Extremely Randomized Trees algorithm, data sampled at 50 Hz and 5-s windows. We conclude by discussing the viability of using IMU sensors to recognize common smartphone activities.

Recommender systems are highly vulnerable to shilling attacks, both by individuals and groups. Attackers who introduce biased ratings in order to affect recommendations, have been shown to negatively affect collaborative filtering (CF) algorithms. Previous research focuses only on the differences between genuine profiles and attack profiles, ignoring the group characteristics in attack profiles. In this paper, we study the use of statistical metrics to detect rating patterns of attackers and group characteristics in attack profiles. Another question is that most existing detecting methods are model specific. Two metrics, Rating Deviation from Mean Agreement (RDMA) and Degree of Similarity with Top Neighbors (DegSim), are used for analyzing rating patterns between malicious profiles and genuine profiles in attack models. Building upon this, we also propose and evaluate a detection structure called RD-TIA for detecting shilling attacks in recommender systems using a statistical approach. In order to detect more complicated attack models, we propose a novel metric called DegSim' based on DegSim. The experimental results show that our detection model based on target item analysis is an effective approach for detecting shilling attacks.