The Predictron: End-to-End Learning and Planning
Abstract
One of the key challenges of artificial intelligence is to learn models that are effective in the context of planning. In this document we introduce the predictron architecture. The predictron consists of a fully abstract model, represented by a Markov reward process, that can be rolled forward multiple “imagined” planning steps. Each forward pass of the predictron accumulates internal rewards and values over multiple planning depths. The predictron is trained end-to-end so as to make these accumulated values accurately approximate the true value function. We applied the predictron to procedurally generated random mazes and a simulator for the game of pool. The predictron yielded significantly more accurate predictions than conventional deep neural network architectures.
Cite
Text
Silver et al. "The Predictron: End-to-End Learning and Planning." International Conference on Machine Learning, 2017.Markdown
[Silver et al. "The Predictron: End-to-End Learning and Planning." International Conference on Machine Learning, 2017.](https://mlanthology.org/icml/2017/silver2017icml-predictron/)BibTeX
@inproceedings{silver2017icml-predictron,
title = {{The Predictron: End-to-End Learning and Planning}},
author = {Silver, David and Hasselt, Hado and Hessel, Matteo and Schaul, Tom and Guez, Arthur and Harley, Tim and Dulac-Arnold, Gabriel and Reichert, David and Rabinowitz, Neil and Barreto, Andre and Degris, Thomas},
booktitle = {International Conference on Machine Learning},
year = {2017},
pages = {3191-3199},
volume = {70},
url = {https://mlanthology.org/icml/2017/silver2017icml-predictron/}
}