Discovering the arrow of time in machine learning

Machine learning (ML) is increasingly useful as data increases in both volume and accessibility. The data inviting ML analysis often includes at least an implicit order or temporal context, which obscures but does not remove the one-directional flow of time within the data. This research takes the first step in exploring the interaction of ML algorithms and training regimes on data with implicit representations. This research will then inform on the suitability of ML for analysing the kind of data that is accumulating daily from every social media platform, Internet of Things device, businesses report, transport tracker or other source. The algorithms are explored first through a literature review and second through an experiment that applies each algorithm to the same data in different ways, each representing time differently. The research is expected to show that ML algorithms are sensitive to temporal context, even when the representation of time in the data or task is only subtle. Further, the research presents preliminary results showing that different training regimes can be understood as ways to represent time within ML, further expanding the set of tools available to researchers when selecting appropriate algorithms.