We know that multimodal-multisensor data is profoundly different from past data sources. It is extremely rich and dense data that typically involves multiple time-synchronized data streams, and it also can be analyzed at multiple levels such as signal, activity pattern, representational, transactional, etc. When multimodal-multisensor data are analysed at multiple levels, they constitute a vast multi-dimensional space for discovering important new phenomena with applied artificial intelligence methods (The Handbook of Multimodal-Multisensor Interfaces, Vol I, https://dl.acm.org/doi/book/10.1145/3015783).
This year's course focusses on Data Analysis Pipelines for Multivariate Time Series for Sustainability: Yearly greenhouse gas emissions of OECD countries, fluctuations on the population size of endangered species, sensor readings on a biochemical reactor: multivariate time series data are generated whenever someone monitors a phenomenon over time. Extracting knowledge from them is a process that starts with obtaining the data, iteratively visualising and transforming, and finally summarising the data into an interpretable representation – whether graphical or mathematical. This course will cover good practices and practical aspects of all steps in the process – handling file input, organising a project’s code, transforming the data with spectral and machine learning methods, and generating models and visualisations that capture relevant structure in the data.
