Proposal of Ephemeral Value Adjustment with Dimensionality Reduction in Deep Reinforcement Learning | IEEE Conference Publication | IEEE Xplore