EEG emotion recognition from facial expression stimuli with the presence of noisy labels

Instructions in the content of each file uploaded to this repository:

DEAP DATASET:

• ICA_DEAP_WITH_RESULTS: Contains the preprocessing pipeline with ICA and results from any regression algorithm, including curves drawn for hyperparameter tuning.
• ICA_DEAP_VISUALS: Contains any plots regarding preprocessing with ICA, such as heatmaps, independent components and EEG before and after ICA.
• VAE_DEAP: Contains the preprocessing pipeline with VAE and results from any regression algorithm, including curves drawn for hyperparameter tuning.

MY DATA:

• ICA_MYDATA_WITH_RESULTS: Contains the preprocessing pipeline with ICA and results from any regression algorithm, including curves drawn for hyperparameter tuning.
• ICA_MYDATA_VISUALS: Contains any plots regarding preprocessing with ICA, such as heatmaps, independent components and EEG before and after ICA.
• VAE_MYDATA: Contains the preprocessing pipeline with VAE and results from any regression algorithm, including curves drawn for hyperparameter tuning.

EXCEL SPREADSHEET:

• MyData.xlsx: Data analysis that was conducted for our dataset, to detect where have participant misclassified labels of valence, arousal and dominance. A few comparisons drawn on the last sheet.

Notes:

The code for the implementation of the VAE was inspired by Li et.al although a lot of changes has been done to the original code.