Topic modeling is an effective tool for analyzing the thematic content of large collections of text. However, traditional probabilistic topic modeling is limited to a small number of topics (typically no more than hundreds). We introduce fine-grained topic models, which have large numbers of nuanced and specific topics. We demonstrate that fine-grained topic models enable use cases not currently possible with current topic modeling techniques, including an automatic cross-referencing task in which short passages of text are linked to other topically related passages. We do so by leveraging anchor methods, a recent class of topic model based on non-negative matrix factorization in which each topic is anchored by a single word. We explore extensions of the anchor algorithm, including tandem anchors, which relaxes the restriction that anchors be formed of single words. By doing so, we are able to produce anchor-based topic models with thousands of fine-grained topics. We also develop metrics for evaluating token level topic assignments and use those metrics to improve the accuracy of fine-grained topic models.

This thesis studies real-time reference ramping of human input for remotely operated vehicles and its effect on system control, power usage, and user experience. The implementation, testing, and evaluation were done on the remotely operated Blueye Pioneer underwater drone. The developed method uses minimum jerk trajectories for transitioning between varying target velocities with a constant end jerk target. It has a low computational cost and runs in real-time on the Blueye Pioneer underwater drone. The presented method produces a well-defined reference with continuous position, velocity, and acceleration states that can be used in the feedback loop. Experiments and simulations show that the method produces a smoother and more predictable motion path for the user. The motions are better suited for video recordings and remote navigation, compared to the direct usage of human input velocity. The smoother reference reduces the controller tracking error, the peak control input, and the energy usage. The introduced acceleration reference state is used for feedforward control on the system. It improves the feeling of controlling the drone by reducing the system lag, the position tracking error, and the rise time for velocity changes.