Jurkus Robertas - Institute of Data Science and Digital Technologies

Doctoral studies

Theme

Maritime Traffic Awareness Evaluation Using Deep Neural Networks

Supervisor Assoc. Prof. Dr. Povilas Treigys

Study time 2020 10 01–2024 09 30

Aim

Investigate deep recurrent neural networks for monitoring the awareness of sea traffic - predicting the movement of a vessel.

Main problems:

Application of recurrent neural network architectures for cargo vessel trajectory prediction.
Study of hyperparameters of recurrent neural networks - determination of cell size.
Application of different coordinate systems in identifying geolocation and their impact on forecasting.
Introduction of different vessel types to the data set using the autoencoder architecture and quartile prediction evaluation.

In 2023:

Developed an article and a poster presentation on the topic: "Categorical Data Encoding Techniques for Recursive Multi-Step Prediction of Vessel Trajectory". International Conference on Time Series and Forecasting (ITISE 2023), held in Gran Canaria, 12-14 July, 2023 (Spain).
Participated in the VU multidisciplinary and international summer PhD school: "Integrity and Visibility: Challenges for Early Career Researchers". 19-22 June 2023.

In 2022:

An article was prepared at the student scientific conference "Technology and Business Current Affairs" and a report was presented "The influence of vessel types on predicting the trajectory of vesel movement using deep recurrent neural networks". KTU, Panevėžys Faculty of Technology and Business.
The presentation was read at the event "AI Lithuania Meetup Klaipėda", the title of the presentation was "Application of polar and Cartesian coordinate systems for trajectory prediction using recurrent neural networks", University of Klaipėda.
Participated in the competition and read the report "The transport theme of students' final theses in the competition". The winner of the prize place, on the topic of water transport, entitled "Study of LSTM deep neural networks for predicting ship progress using big traffic data", Ministry of Communications, Vilnius, Lithuania.
Participated in the event "Baltic Sea Science Day Award for Young Scientists 2022", research area: "Digital growth", topic: "Maritime Traffic Awareness Evaluation Using Deep Neural Networks", University of Latvia, Norwegian Presidency 2021-2022.

In 2021:

A poster presentation was prepared, on the topic: "Prediction of vessels trajectory using different coordinate systems". DAMSS: 12th conference on data analysis methods for software systems, Druskininkai, Lithuania.
The article was presented at the conference "Investigation of recurrent neural network architectures for prediction of vessel trajectory". Information and software technologies: 27th international conference, ICIST 2021, Kaunas, Lithuania.

In 2016:

Participated in the international scientific-practical student conference and read the report "Modeling of Information Systems and Computer Equipment and Household Appliance Stores Business".
Participated in the student conference and read the report "Incident Data Analysis Cube".