Training Problem Sample Clauses
Training Problem. In Chapter 2, we introduce a training problem in a deterministic, discrete- timed, and finite reinforcement learning system. In this chapter, we intend to numerically solve the previous problem and compare the results we attained by applying the linear program approach (3.14) formulated in Chapter 3.
Training Problem. Just as the problem stated in Chapter 2, suppose we have a determinis- tic and finite reinforcement learning problem. Let the state space be X = {0, 1, 2, 3, 4, 5, 6} and at each state we have the actions U = {−1, 0, 1} and the model is
1) The running cost is given by c(x, u) = 0 x = 0 or x = 6 1 else. Let the distribution of initial states, ρ, be defined by 5 ρ(x) = 1 , x ∈ {1, 2, 3, 4, 5} (4.3) 0 else.