Spatiotemporal networks (networks whose topology and attributes change over time) play an important role in modern computer science – for modelling different types of real-life relationships (computer networks, social networks, etc.), for representation of rich data structures like texts or images, or just as a brick for modern model architectures like graph convolutional neural networks. This subsession is focused on data-driven learning of spatiotemporal network structures and discusses existing challenges and emerging approaches.

Data Science Fundamentals: this subsession is focused on application of machine learning models to solve real world problems, for example Markov-modulated linear regression and its application in transportation. It also addresses modelling, traffic simulation and examples of different real-world projects.  

1. Spatiotemporal networks: definition and applications

• Data-driven learning of network structures: principles and approaches
• Links between learning of network structures and multivariate time series analysis and feature selection
• Algorithms of network structures’ learning: correlation-based, regularization-based, mutual information-based, homogeneous and heterogeneous ensembles
• Application: learning spatiotemporal relationships in traffic flows

2. Data science fundamentals:

• Introduction to core concepts and technologies: the data science process, toolkit and types of data.
• Predictive Modeling, Model Fitting, and Regression Analysis.
• Markov-modulated linear regression and its application in transportation (postdoc project).

3. Mathematical modeling. Simulation.

• Analytical versus algorithmic.
• TSI experience: traffic modeling, examples of the real-world projects.

Dr Dmitry Pavlyuk is a professor at Transport and Telecommunication Institute (TSI), Latvia. He holds a master’s degree in computer science and two doctoral degrees – in economics (2005) and in civil engineering (2015). Dmitry is the author of more than 50 scientific publications, including 37 indexed in Scopus that have more than 160 citations. His research interests are mainly focused on spatial econometrics, spatiotemporal big data modelling and machine learning in transportation.

Dr.Sc.Ing. Nadezda Spiridovska is an associate professor and a researcher of Engineering Faculty at Transport and Telecommunication Institute. Nadezda Spiridovska defended Doctor Degree in Engineering (Doctoral Degree Programme "Telematics and Logistics") in 2015 in Transport and Telecommunication Institute. Scientific interest is related to transport modelling, data science, simulation. Nadezda is the author of more than 20 scientific publications (cited by international scientific publishers).

Sesión única de presentación en modalidad híbrida (presencial y remota) con turno de debates y preguntas al final

Asistencia presencial o a través de Collaborate (con informe de conexión) y superar la prueba de preguntas sobre la presentación disponible en el correspondiente curso de Aula Virtual