PhD in artificial intelligence for segmentation of time series
AI systems often have to cope with an overwhelming amount of input data. Time series often contain long sequences which have a similar behavior and which can be processed as a whole, rather than cutting it in a large amount of fixed-size epochs. In this PhD project, we will automate such segmentation processes as much as possible, with a focus on automatically splitting (multi-modal) time series into segments of variable sizes, within which the statistics are homogeneous across each segment. Such segmentation allows to model or process each segment with a different (more tailored) model, or to treat each segment as a higher-level object, which can be described or embedded as a single feature vector in further processing steps.
The main goal is to design a general-purpose segmentation pipeline and apply it in several use cases, for example in the analysis of electroencephalography (EEG) data for epilepsy.