This repository contains the code and data to reproduce the findings from our paper.
All wrappers can be found in the 'examples' folder:
test_dynamic_task4.py is the wrapper to train a reinforcement learning agent.
test_dynamic_task_sup is the wrapper to train a supervised agent.
Load_task4_final generates all reinfrocement learning figures from the paper.
Load_supervised_final generates all supervised figures from the paper.
Code is written in Python 2.7. Neural networks are based on the Chainer library [1]
Required libraries:
- Chainer V1
- Numpy
- Scipy
- Matplotlib
- time
- scikit-learn
[1] Tokui, S., Oono, K., Hido, S. and Clayton, J., Chainer: a Next-Generation Open Source Framework for Deep Learning, Proceedings of Workshop on Machine Learning Systems(LearningSys) in The Twenty-ninth Annual Conference on Neural Information Processing Systems (NIPS), (2015)