LWRP: Low Power Consumption Weighting Replacement Policy using Buffer Memory

While off-policy temporal difference (TD) methods have widely been used in reinforcement learning due to their efficiency and simple implementation, their Bayesian counterparts have not been utilized as frequently. One reason is that the non-linear max operation in the Bellman optimality equation makes it difficult to define conjugate distributions over the value functions. In this paper, we introduce a novel Bayesian approach to off-policy TD methods, called as ADFQ, which updates beliefs on state-action values, Q, through an online Bayesian inference method known as Assumed Density Filtering. We formulate an efficient closedform solution for the value update by approximately estimating analytic parameters of the posterior of the Q-beliefs. Uncertainty measures in the beliefs not only are used in exploration but also provide a natural regularization for the value update considering all next available actions. ADFQ converges to Q-learning as the uncertainty measures of the Q-beliefs decrease and improves common drawbacks of other Bayesian RL algorithms such as computational complexity. We extend ADFQ with a neural network. Our empirical results demonstrate that ADFQ outperforms comparable algorithms on various Atari 2600 games, with drastic improvements in highly stochastic domains or domains with a large action space.

show abstract

LWRP: Low Power Consumption Weighting Replacement Policy using Buffer Memory

Cited by 3 publications

References 7 publications

Enhanced Not Recently Used Algorithm for Cache Memory Systems in Mobile Computing

Enhanced Not Recently Used Algorithm for Cache Memory Systems in Mobile Computing

Intermediate data placement and cache replacement strategy under Spark platform

Assumed Density Filtering Q-learning

Contact Info

Product

Resources

About