TY - JOUR

T1 - Ontogenetic and Phylogenetic Reinforcement Learning

AU - Togelius, Julian

AU - Schaul, Tom

AU - Wierstra, Daan

AU - Igel, Christian

AU - Gomez, Faustino

AU - Schmidhuber, Juergen

PY - 2009

Y1 - 2009

N2 - Reinforcement learning (RL) problems come in many ﬂavours, as do algorithms for solving them. It is currently not clear which of the commonly used RL benchmarks best measure an algorithm’s capacity for solving real-world problems. Similarly, it is not clear which types of RL algorithms are best suited to solve which kinds of RL problems. Here we present some dimensions along the axes of which RL problems and algorithms can be varied to help distinguish them from each other. Based on results and arguments in the literature, we present some conjectures as to what algorithms should work best for particular types of problems, and argue that tunable RL benchmarks are needed in order to further understand the capabilities of RL algorithms

AB - Reinforcement learning (RL) problems come in many ﬂavours, as do algorithms for solving them. It is currently not clear which of the commonly used RL benchmarks best measure an algorithm’s capacity for solving real-world problems. Similarly, it is not clear which types of RL algorithms are best suited to solve which kinds of RL problems. Here we present some dimensions along the axes of which RL problems and algorithms can be varied to help distinguish them from each other. Based on results and arguments in the literature, we present some conjectures as to what algorithms should work best for particular types of problems, and argue that tunable RL benchmarks are needed in order to further understand the capabilities of RL algorithms

KW - -Reinforcement Learning (RL)

KW - -Algorithm Benchmarks

KW - -RL Problem Dimensions

KW - -Real-World Problem Solving

KW - -Tunable RL Benchmarks

M3 - Journal article

SN - 0933-1875

VL - 2009

JO - KI - Künstliche Intelligenz

JF - KI - Künstliche Intelligenz

IS - 3

ER -