Artificial Intelligence By Example

Implementing Z – squashing the MDP result matrix

The output of the MDP functions provides the following Q matrix:

Q :[[ 0. 0. 0. 0. 321.8 0. ] [ 0. 0. 0. 401. 0. 258.44] [ 0. 0. 0. 401. 0. 0. ] [ 0. 0. 0. 500. 0. 0. ] [ 258.44 0. 0. 401. 0. 0. ] [ 0. 321.8 0. 0. 0. 0. ]]

Each line represents a vertex in the graph: A, B, C, D, E, and F. Each value obtained needs to be squashed in the following z(x) function:

The first step in the code that follows is to squash the weights provided by the MDP process for each line (vertex) x with a logistic_sigmoid function:

119:#Logistic Sigmoid function to squash the weights def logistic_sigmoid(w):return ...