Overcomplete sparse coding of natural images enables reinforcement learning to solve optimal control tasks orders of magnitude larger than with complete codes, via a new scalable benchmark and theoretical justification.
Emergence of simple-cell receptive field properties by learning a sparse code for natural images
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Optimal Control with Natural Images: Efficient Reinforcement Learning using Overcomplete Sparse Codes
Overcomplete sparse coding of natural images enables reinforcement learning to solve optimal control tasks orders of magnitude larger than with complete codes, via a new scalable benchmark and theoretical justification.