deletions | additions       diff --git a/section_Introduction_The_occupancy_map__.tex b/section_Introduction_The_occupancy_map__.tex  index 5aaee76..b80ffb0 100644  --- a/section_Introduction_The_occupancy_map__.tex  +++ b/section_Introduction_The_occupancy_map__.tex 
...
\section{Introduction}  The occupancy map problem learns a probabilistic model of some space. We map every point in the space to the probability that said space occupied. For a primer on the problem, see \cite{ramoshilbert}. Recently, a new approach for solving this problem in an effective and on-line manner was proposed. We use a logistic regression model to assign probabilities. Linear separability and model applicability is ensured through the use the kernel trick. We map the low-dimensional feature vector to a much denser space and fit a linear classification algorithm to it. Furthermore, kernel approximation methods are employed to ensure low run-time.  Such a model has been demonstrated to perform very well in terms of accuracy/time tradeoff \cite{ramoshilbert}. It does however suffer from a dependence on a number of hyperparameters. Solving this problem would allow for better results and a wider application.  To this end, we first explore a model-shift from frequentist to Bayesian logistic regression formulation. Our goal is to both determine whether we can get more accurate models. Furthermore, we seek to determine if the paradigmatic shift in our characterisation of uncertainty means we're less reliant on the hyperparameters of the model.  Our next focus is on Bayesian parameter optimisation. We attempt to tackle the hyper-parameter problem more directly through the use of intelligent automatic selection algorithms.