@inproceedings{7589da58bb1a485a82b4f750703bd632,
title = "Intelligence as inference or forcing Occam on the world",
abstract = "We propose to perform the optimization task of Universal Artificial Intelligence (UAI) through learning a reference machine on which good programs are short. Further, we also acknowledge that the choice of reference machine that the UAI objective is based on is arbitrary and, therefore, we learn a suitable machine for the environment we are in. This is based on viewing Occam's razor as an imperative instead of as a proposition about the world. Since this principle cannot be true for all reference machines, we need to find a machine that makes the principle true. We both want good policies and the environment to have short implementations on the machine. Such a machine is learnt iteratively through a procedure that generalizes the principle underlying the Expectation-Maximization algorithm.",
author = "Peter Sunehag and Marcus Hutter",
year = "2014",
doi = "10.1007/978-3-319-09274-4_18",
language = "English",
isbn = "9783319092737",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "186--195",
booktitle = "Artificial General Intelligence - 7th International Conference, AGI 2014, Proceedings",
address = "Germany",
note = "7th International Conference on Artificial General Intelligence, AGI 2014 ; Conference date: 01-08-2014 Through 04-08-2014",
}