Laboratory of Machine Learning in Earth Sciences
Head of Laboratory
Krinitskiy, Mikhail V
Publications
44
Citations
150
h-index
7
Authorization required.
In the laboratory, we are engaged in fundamental and applied research in the fields of geophysics, oceanology, marine geology, marine biology, meteorology, ocean-atmosphere interaction, and climate dynamics. The research is carried out using modern and classical machine learning methods, including artificial neural networks. Among the tasks that are solved in the laboratory, it is possible to list monitoring, measurements, conducting and maintaining field observations, as well as modeling natural processes at various scales. The initial data can be in situ observational data, data from remote sensing of the Earth from space, from unmanned or manned aircraft, as well as data from geophysical modeling.
- Machine learning
- Artificial intelligence
- Statistics
- Mathematical modeling
- Mathematical and physical modeling
- Mathematical statistics
- Collection and analysis of geophysical data
- Pattern recognition methods for studying the Earth's magnetic field and solving other geophysical problems
- Analysis of Earth remote sensing data
- Remote sensing Data (DDZ)
Mikhail Krinitskiy
Head of Laboratory
Viktor Golikov
Researcher
Vadim Rezvov
Researcher
Olyga Belousova
Engineer
Mihail Borisov
Junior researcher
Research directions
Analysis of ship navigation radar images
+
In this study, we train machine learning models to determine the characteristics of wind waves based on backscattering amplitude data recorded by the navigation radar of marine vessels.
Identification and classification of floating marine debris in the ocean
+
In this study, we train artificial neural networks to detect and classify floating marine debris and other objects atypical of the sea surface, based on high-resolution optical imaging from a marine vessel.
Analysis of the characteristics of the return migration of Pacific salmon
+
In this project, we train statistical models to predict the date of the median return and the share of the northern division for the return migration of Pacific salmon to the Fraser River (Canada)
Statistical downscaling and correction of ocean and atmosphere models
+
In this research, we train artificial neural networks to enhance resolution (a process known as "downscaling") or for statistical correcction of the results of atmospheric and oceanic dynamics modeling. The most important problem being addressed in this research is the development of methods for assessing the quality of neural network approaches in downscaling and statistical correction.
Publications and patents
Lab address
Долгопрудный, Институтский переулок, 9
Authorization required.