Title: In-situ Semantic Segmentation of Streaming Time-series Data
Committee:
Dr. Anderson, Advisor
Dr. Dyer, Chair
Dr. Inan
Abstract: The objective of the proposed research is to build a tool that notices key observations for scientists. It will semantically segment heterogeneous, multi-channel streaming data in-situ in a resource-constrained package and quickly present it in a useful manner. We will investigate metrics for this tool that do not rely on ground truth labels. Finally, we will present a constrained proof-of-concept “live” demonstration. Ultimately, this will be released as a tool for exploration that facilitates the study of how the best human learners learn in-situ.