Reinforcement learning for real-world tasks

The book is dedicated to the industrial-oriented use of training with reinforcement (Reinforced learning, RL) It is explained how to teach industrial and scientific systems to solve any step -by -step problems by trial and error - without preparing highly specialized educational sets of data and without risk to re -rendering or replacing the algorithm. Markov decision-making processes, deep Q-networks, gradients of politicians and their calculation, methods for eliminating entropy and much more are considered. This book is the first in Russian, where the theoretical basis RL and the algorithms are given in applied, industry manner.
For analysts of data and specialists in artificial intelligence