4IAR-nns

Implementation of the deep neural networks described in Kuperwajs, Schütt, and Ma (2023).

Approach

This repository implements networks as models of human play in 4-in-a-row using PyTorch. The networks are trained and tested on human decisions in large-scale data. Trained networks are available upon request, and the cognitive model code is available at https://github.com/ionatankuperwajs/4IAR-improvements.

File description

• network.py: architecture for the neural networks
• preprocessing.py and custom_dataset.py: data formatting
• load_train.py and training.py: training scripts
• load_test.py and testing.py: testing scripts
• analysis.py and summary_stats.py: analysis scripts for figures in the paper
• model_preprocessing.py, model_comparison.py, and model_improvements.py: data formatting and analysis for the cognitive model comparison
• train_network.sh, train_network_array.sh, and test_network.sh: bash scripts for training and testing on the computing cluster