DETECTING ANOMALIES IN INTERMAGNET DATA USING GRAPH NEURAL NETWORK
Annotation
The use of modern digital information technologies, such as Data Mining, Data Science and Big Data, has caused an exponential growth in the volume of data, allowing to obtain new knowledge in various subject areas based on the information provided. In this regard, tasks related to pre-processing, intellectual analysis, visualization of large data sets have become especially relevant. Using the methods of intelligent analysis Unsupervised learning, the problem of detecting anomalies (outliers) in data arrays obtained from the Lycksele magnetic observatory, which is part of the international network INTERMAGNET, is solved. Since anomalies reflect changes in the Earth’s geomagnetic field, they are highly informative, which gives the solution to this problem great scientific and practical value. Anomalies in the designated data are not frequent enough, so they can be detected only in a large volume of processed information. The results of detecting anomalies using a graph neural network are presented. The MATLAB system is used as a software tool.
Keywords
Постоянный URL
Articles in current issue
- CHANNEL ERROR OF THE AIR MOTION PARAMETERS MEASUREMENT SYSTEM FOR AIRCRAFT WITH AN INTEGRATED FUSELAGE FLOW RECEIVER
- MODELING OF TRANSVERSE TANGENTIAL INTERACTION OF THE FOOT WITH THE SUPPORTING SURFACE
- MODEL ASSESSMENT OF THE POSSIBILITY OF USING A STEPPED TRANSMISSION IN AN ELECTRIC VEHICLE
- ALGORITHM FOR REDISTRIBUTING ROLES IN WIRELESS COMPUTING CLUSTERS
- INFLUENCE OF INITIATOR MATERIAL ON THE PROCESS OF OPTICAL BREAKDOWN IN OPTICAL FIBER
- DEVELOPMENT OF A DIAGNOSTIC SYSTEM FOR A HYBRID POWER PLANT OF A LIGHT AIRCRAFT
- USE OF POLYMER COMPOSITE MATERIALS IN SPUR GEAR
- CENTERLESS CONTROL SCHEME FOR LARGE-SIZED SHELL PROFILES DURING THEIR FORMATION
- VARIABLE IMPEDANCE LEARNING CONTROL FOR ROBOTIC ARMS FROM GMR-ENCODED BEHAVIOR PRIORS
- SPECTROPHOTOMETRIC METHOD FOR STUDYING COMPOSITES CONTAINING METHACRYLIC MONOMERS