Bayesian Inference for Linear Models and Decision Making
This project was completed for AEROSP 567: Statistical Inference, Estimation and Learning at the University of Michigan.
Relevant Skills and Topics:
Python
Gaussian Processes
Lineat GPR
Hyperparameter Tuning and Kernel Design
Given a source for where a signal is coming from with sparse data, determine the location of the source using Gaussian Processes. The process involved guessing the location of the source 12 times total with 3 different requests.
Determining the source was a trade-off between exploration and exploitation of the sparsely represented search space, where the squared exponential kernel was used for the covariance metric and a tunable exploration vs. exploitation metric was used in order to select the next data points.
Uncertainty quantification was then evaluated using Confidence Intervals.