Predicting Word Stimulus from fMRI Neural Activations: Gaussian Naive Bayes Classifier

Studies have shown that thinking about different semantic categories of words (for example, tools, buildings,and animals) activates different spatial patterns of neural activation in the brain.

Given an fMRI image of 21764 voxels, this Gaussian Naive Bayes classifier predicts the associated stimulus word/class given to a human subject based on their observed neural activity measured by functional magnetic resonance imaging (fMRI).

References

This was developed for course 10-601 Machine Learning at Carnegie Mellon University in Spring 2021, taught by Professor Tom Mitchell and Matt Gormley.

The data and experiments originated from a research paper from Professor Tom Mitchell et al: "Predicting Human Brain Activity Associated with the Meanings of Nouns" For more information visit http://www.cs.cmu.edu/~tom/science2008/index.html.

Usage

To run the Gaussian Naive Bayes classifier:

python3 gnb.py train_data.csv test_data.csv train_out.labels test_out.labels

To visualize the 2D slices of the brain:

python3 visualize.py <path/to/dataset> <row_index_into_dataset>