A position paper arguing that deep RL should pursue 'fluid adaptivity' through modular, dynamically interacting control structures inspired by biological systems, rather than static behavioral-space representations.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
From Crystallized Adaptivity to Fluid Adaptivity in Deep Reinforcement Learning -- Insights from Biological Systems on Adaptive Flexibility
A position paper arguing that deep RL should pursue 'fluid adaptivity' through modular, dynamically interacting control structures inspired by biological systems, rather than static behavioral-space representations.