Sequential Source Density Estimation

This repository develops a sequential extension of the source density estimation algorithm Sourcerer.

The main algorithm is developed in notebooks/benchmark_sequential_naive_script.py for the benchmark simulators, and in notebooks/hh_sequential_naive_script.py for the Hodgkin-Huxley simulator.

The launching scripts are scripts/launch_jobs.sh and scripts/launch_jobs_hh.sh.

All results and figures can be found in results_analysis/ directory. Figures for the benchmark simulators are in the notebook naive_seq_benchmark_results_25_11.ipynb and Hodgkin-Huxley results are available in naive_seq_benchmark_results_hh.ipynb.

Sourcerer

This is the repository for the paper "Sourcerer: Sample-based Maximum Entropy Source Distribution Estimation".

Installation

Install all dependencies with pip install -e .

Running TM, IK, SLCP, GM or SIR, Lotka-Volterra tasks:

hydra is used for configuration management and command line interface.

You can run all benchmark tasks either from the command line or interactively (with VSCode cells or Jupyter notebooks) using the benchmark_simulator_script.

To run them interactively, overwrite the local_overrides list with your desired configuration.

Alternatively, you can run the script from the command line: python3 benchmark_simulator_script.py simulator=two_moons (as an example, the two_moons task is selected here).

By default, results will be saved in results_sourcerer.

Running the Hodgkin-Huxley experiment

To run the Hodgkin-Huxley experiment, the public dataset from Scala et al. (2020) is required. The dataset can be downloaded from https://dandiarchive.org/dandiset/000008/draft.

Alternatively, a preprocessed dataset along with a set of simulations to train the surrogate is publicly available at https://github.com/berenslab/hh_sbi.

To train the surrogate, use the hh_sims_and_stats script. To perform source estimation after training the surrogate, use the hh_script. Both can be run interactively using VSCode cells or Jupyter notebooks.

Figures

Code for the figures can be found in the figures folder. To reproduce the figures, it is necessary to first run the experiments and update the paths to point to the result files.

References

Parts of this repository are based on code from the following public repositories:

• https://github.com/berenslab/hh_sbi
• https://github.com/MaximeVandegar/NEB

Citation

@inproceedings{
  vetter2024sourcerer,
  title={Sourcerer: Sample-based Maximum Entropy Source Distribution Estimation},
  author={Julius Vetter and Guy Moss and Cornelius Schr{"o}der and Richard Gao and Jakob H. Macke},
  booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems},
  year={2024},
  url={https://openreview.net/forum?id=0cgDDa4OFr}
}