NeuroCERIL

neuroceril includes an implementation of NeuroCERIL, a programmable neural network that implements a causal inference algorithm for imitation learning

Davis, G.P., Katz, G.E., Gentili, R.J., Reggia, J.A. NeuroCERIL: Robotic Imitation Learning via Hierarchical Cause-Effect Reasoning in Programmable Attractor Neural Networks. Int J of Soc Robotics 15, 1277–1295 (2023). publication preprint

Requirements

• numpy is required for the base implementation
• pycuda is required to run on GPUs
• networkx is required to generate unification test cases
• matplotlib is required to use built-in plotting of test data

Installation

1. Clone or download this repository into a directory of your choice.
2. Add the src sub-directory to your PYTHONPATH.

Basic Usage

The build_neuroceril.py script contains functions for constructing a NeuroCERIL instance and loading it with imitation learning programs. The test_neuroneuroceril.py includes an implementation of the causal inference algorithm, and a test function that loads a causal knowledge base and test demonstration. Causal knowledge base and demonstration files are contained in the kb and demos directories, respectively.

Example command to run the replace_red_with_spare_1 demonstration:

python3 test_neuroceril.py -t imitate -k kb_il_streamlined -n demo_replace_red_with_spare_1 -v

This will run the full neural implementation, which is quite computationally expensive. To emulate the neural architecture, use the -m flag:

python3 test_neuroceril.py -t imitate -k kb_il_streamlined -n demo_replace_red_with_spare_1 -v -m