SCS Undergraduate Thesis Topics
|Sunsern Cheamanunkul||Tom Mitchell/Rebecca Hutchinson||Gaussian Naive Bayes Classifier with Smooth Basis Functions|
We explore another way to identify human mental states using the data from fMRI machine. An fMRI machine measures the changes in blood oxygenation level, which is also known as hemodynamic responses (HDR). Since a hemodynamic response corresponds to a blood flow, it is convincing that fMRI data should have smoothness property. In this thesis, we extend the traditional Gaussian Naive Bayes classifier by adding smooth basis functions into the calculation to capture the smoothness property of fMRI data. After some experiments, we do not see a statistically significant improvement by using basis functions. However, we investigate further into the questions why using basis functions did not help and in what cases where using basis functions could improve the classification.