Influence Networks Algorithms: Finding the best decision

From DSL

Sometimes the only thing that we might be interested in is the optimal decision at the top level of the influence diagram. Since the policy evaluation algorithm computes the expected utilities of all possible policies, it may be doing unnecessary work. For those cases, the SMILE library and GeNIe provide a simplified but very fast algorithm written by Shachter and Peot (1992). The algorithm instantiates the first decision node to the optimal decision alternative but does not produce the numerical expected utility of this or any other decision option. In order for this algorithm to be run, all informational predecessors of the first decision node have to be instantiated.

Let us consider an example.

Shown below is a simple Influence Diagram from Tutorial 4 - Creating Influence Diagrams.

After selecting 'Find Best Policy' from Network Menu and updating the network the following will be visible.

As you can see, only the Decision node has been updated. Also it isn't updated with the actual utilities, rather, it only tells which decision is the best decision [ which decision would yield the highest utility ] . If we had other decision nodes in the network, they wouldn't be updated at this stage. This is because the 'Find Best Policy' algorithm only finds the best policy for the 'highest' decision node in the network [ one that doesn't have any decision nodes before it in temporal order ] To find the best policy for other decision nodes in the network, you first need to set the decision for the first decision node [ in our case, Investment Decision to say DonotInvest ] and re update the network. Now, the algorithm will find the best policy for the next decision node in temporal order and so on.

If you need to see the actual utlities for the decisions the Policy Evaluation algorithm is a better choice.