Public defence in Computer Science, M.Sc. Valerii Iakovlev

Title of the thesis: Deep Learning Methods for Modeling of Spatiotemporal Dynamical Systems Governed by Partial Differential Equations

Doctoral student: Valerii Iakovlev
Opponent: Associate Professor Melih Kandemir, University of Southern Denmark, Denmark
Custos: Professor Harri Lähdesmäki, Aalto University School of Science, Department of Computer Science

The problem of teaching computers to understand and predict complex processes that change over space and time is addressed in this dissertation. Spatiotemporal systems, which include applications ranging from weather forecasting to traffic analysis, are ubiquitous in everyday life. However, the training of deep learning models for such systems is often hindered by significant limitations, including the need for extensive datasets, reliance on structured and evenly spaced measurements, and instability during training. 

In this dissertation, new methods are developed to overcome these limitations. The developed methods enable the modeling of more complex systems, where data may be sparse, irregular, or incomplete, and where fast, reliable training is required. These advancements address limitations of existing deep learning methods for spatiotemporal modeling, bringing deep learning-based solutions closer to real-world applications. Applications in fields such as climate science, transportation, and biology stand to benefit from these improvements, leading to better predictions, insights, and decision-making. 

Thesis available for public display 10 days prior to the defence at: https://aaltodoc.aalto.fi/doc_public/eonly/riiputus/ 

Contact information:

Doctoral theses of the School of Science: https://aaltodoc.aalto.fi/handle/123456789/52