Smartwatch-based semantic learning of elements of human behaviour

Thesis

The modern interconnected world is characterised by a plethora of mobile applications that operate in smart mobile and wearable devices, aiming to assist various aspects of human life, such as smart home operations or remote monitoring of older adults. Understanding daily patterns of human behaviour is key to providing informed and robust solutions in such scenarios. By inspecting the focus of popular mobile applications closely it becomes apparent that they commonly share three main requi...

Expand abstract

Authors: Alevizaki, A

• DOI: 10.5287/ora-ovz6x06gr
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: human activity recognition; dataset cleaning; model generalisation; data alignment; hierarchical learning; domain adaptation; data collection; indoor positioning; transfer learning; indoor localisation
• Subjects: Model Generalisation; Transfer Learning; Dataset Cleaning; Deep learning (Machine learning); Data sets; Human activity recognition; Indoor positioning systems (Wireless localization); Machine learning; Data Alignment