We describe an expressive class of policies that can be efficiently learned
from a few demonstrations. Policies are represented as logical combinations of
programs drawn from a small domain-specific language (DSL). We define a prior
over policies with a probabilistic grammar and derive an approximate Bayesian
inference algorithm to learn policies from demonstrations. In experiments, we
study five strategy games played on a 2D grid with one shared DSL. After a few
demonstrations of each game, the inferred policies generalize to new game
instances that differ substantially from the demonstrations. We argue that the
proposed method is an apt choice for policy learning tasks that have scarce
training data and feature significant, structured variation between task
instances.
Description
[1904.06317] Few-Shot Bayesian Imitation Learning with Logic over Programs
%0 Journal Article
%1 silver2019fewshot
%A Silver, Tom
%A Allen, Kelsey R.
%A Lew, Alex K.
%A Kaelbling, Leslie Pack
%A Tenenbaum, Josh
%D 2019
%K approximate bayesian few-shot learning
%T Few-Shot Bayesian Imitation Learning with Logic over Programs
%U http://arxiv.org/abs/1904.06317
%X We describe an expressive class of policies that can be efficiently learned
from a few demonstrations. Policies are represented as logical combinations of
programs drawn from a small domain-specific language (DSL). We define a prior
over policies with a probabilistic grammar and derive an approximate Bayesian
inference algorithm to learn policies from demonstrations. In experiments, we
study five strategy games played on a 2D grid with one shared DSL. After a few
demonstrations of each game, the inferred policies generalize to new game
instances that differ substantially from the demonstrations. We argue that the
proposed method is an apt choice for policy learning tasks that have scarce
training data and feature significant, structured variation between task
instances.
@article{silver2019fewshot,
abstract = {We describe an expressive class of policies that can be efficiently learned
from a few demonstrations. Policies are represented as logical combinations of
programs drawn from a small domain-specific language (DSL). We define a prior
over policies with a probabilistic grammar and derive an approximate Bayesian
inference algorithm to learn policies from demonstrations. In experiments, we
study five strategy games played on a 2D grid with one shared DSL. After a few
demonstrations of each game, the inferred policies generalize to new game
instances that differ substantially from the demonstrations. We argue that the
proposed method is an apt choice for policy learning tasks that have scarce
training data and feature significant, structured variation between task
instances.},
added-at = {2019-06-29T22:27:00.000+0200},
author = {Silver, Tom and Allen, Kelsey R. and Lew, Alex K. and Kaelbling, Leslie Pack and Tenenbaum, Josh},
biburl = {https://www.bibsonomy.org/bibtex/27930082dbbebbdeadafdd54924c14b57/kirk86},
description = {[1904.06317] Few-Shot Bayesian Imitation Learning with Logic over Programs},
interhash = {5e3a473f3893fb08604ae6bbe81953ef},
intrahash = {7930082dbbebbdeadafdd54924c14b57},
keywords = {approximate bayesian few-shot learning},
note = {cite arxiv:1904.06317},
timestamp = {2019-06-29T22:27:00.000+0200},
title = {Few-Shot Bayesian Imitation Learning with Logic over Programs},
url = {http://arxiv.org/abs/1904.06317},
year = 2019
}