EAGE Subsurface Intelligence Workshop 2019 9-10 December 2019, Manama, Bahrain POS02 Seismic horizon detection using Convolutional Neural Networks D. Mylzenova 1 *, S. Tsimfer 1 , A. Koryagin 1 , R. Khudorozhkov 1 , S. Zaytcev 1 1 Gazprom Neft Summary This article extends the category of artificial intelligence algorithms by applying Deep Reinforcement Learning (DRL) for reservoir production optimization based on NVP objective. This work concerns with waterflooding optimization problem to maximize the economic returned by intelligently adjusting the injection rates. Also, allowing the DRL system to observe the raw data as a distribution of primary variables (pressure and saturation).