Prospective Learning in Retrospect

Overview

This repository contains the code to reproduce the experiments from our AGI 2025 paper, Prospective Learning in Retrospect.

Dependencies

To setup the conda environment, run

conda env create -f environment.yml

Guide

Run the following self-explanatory notebooks to reproduce the results and figures from the paper. For foraging experiments, the data can be found and downloaded here.

Fig 2: 1-effect_of_sparsity.ipynb

Fig 3: 2_effect_of_time_embeddings.ipynb

Fig 4: 3-online_prospective_learning.ipynb

Fig 5: 4-prospective_forests.ipynb

Fig 7: 7-plot_foraging_results.ipynb

Try the foraging game by running play_foraging.py

Cite us

If you find this code and our paper useful consider citing

will be updated once the paper is on arXiv