Towards Addressing Label Ambiguity in Sequential Emotional Responses Through Distribution Learning

This work highlights the challenge of labelling data with single-label categories, as there may be ambiguity in the assigned labels. This ambiguity arises when a data sample, which can be influenced by previous affective events is labelled with a single-label category (known as priming). Label distribution learning (LDL) is proposed as an approach to contend with the ambiguity among labels.

This repository contains the code to produce the results presented in this work.

Usage

Run cross-trial experiment

The cross-trial experiment is run sequentially for a list of subjects specified in the script run_cross_trial.py. Modify the variables: learn_type and dataset to specify the learning type (either single label or label distribution learning) and the dataset (either SEED or SEED5) respectively.

Once these variables are set, just run the following command:

python run_cross_trial.py

Evaluate cross-trial experiment

To evaluate, i.e. get metrics, of the cross-trial experiment, set again the variables learn_type and dataset and run the following command:

python get_results.py

Please, if you use the code provided and found relevant the work presented in this paper, cite the following reference in your work: