Ipek et al. [17] propose the use of reinforcement learning (RL) [32] to design high-performance self-optimizing memory schedulers. Reinforcement learning works by interacting
with the environment and learning automatically with experience to pick the actions that maximize a desired long-term objective function. Ipek et al. show that, when used to target performance, this approach can outperform existing ad hoc designs by a significant margin.
.