Decoding Reading Goals from Eye Movements

This project aims to decode reading goals (i.e. information-seeking versus ordinary reading) from eye movements using machine learning techniques.

Getting Started

Prerequisites

• Mamba or Conda

Setup

1. Clone the Repository

   Start by cloning the repository to your local machine:

   git clone https://github.com/lacclab/Goal-Decoding-from-Eye-Movements.git
   cd Goal-Decoding-from-Eye-Movements

2. Create a Virtual Environment

   Create a new virtual environment using Mamba (or Conda) and install the dependencies:

   mamba env create -f environment.yaml

Reproducing the results

1. To train the models including the full hyperparameter sweep run bash scripts/sweep_wrapper.sh. This creates sweep configuration files. Run the created files in the terminal.

2. Then, to get the predictions on the test sets run bash scripts/eval_wrapper.sh.

3. To aggregate and display the results run the notebooks/display_results_task_decoding.ipynb notebook.

4. For the error analysis plots run notebooks/error_analysis.ipynb and for the statistical tests stats.ipynb.