tinyML Asia 2021 Flora Salim: Learning compact representation with less (labelled) data from sensors

tinyML Asia 2021 Session: Frameworks, Tools, tinyML for Good Learning compact representation with less (labelled) data from sensors Flora SALIM , Professor, RMIT University, Melbourne, Australia The proliferation of sensors and the Internet of Things leads to new opportunities and challenges for modeling human behaviors. However, most representation learning techniques require a large amount of well-labeled training sets to achieve high performance. Due to the high expense of labeling human and/or system behaviors, approaches that require minimal to no labeled data are becoming more favorable. This motivated us to explore techniques that are data-efficient learning techniques to achieve efficient and compact representations. Approaches including domain adaptation (with minimal data) and pretraining (without labeled data) will be introduced.