pqn
Simplifying Deep Temporal Difference Learning
tl;dr The authors propose PQN, a simplified deep online Q-Learning that uses very small replay buffers. Normalization and parallelized sampling from vectorized …





