Home | Trees | Indices | Help |
|
---|
|
generic.Policy --+ | LookaheadPolicy.LookaheadPolicy --+ | StochasticLookahead
A nondeterministic version of the lookahead-based policy, where actions are selected with a probability that is a function of their expected values.
|
|||
|
|||
|
|||
Inherited from |
|
|||
beta = 1.0
|
|
|||
Inherited from |
|
Returns a randomly selected action out of the available choices, with each action selected with a probability dependent on its relative expected value
|
Computes a probability distribution over the provided dictionary of action choices. Each value in the dictionary must have a 'value' field containing a float. This method computes a Boltzmann distribution based on these values and stores it in the 'probability' field of each entry. Modify the 'beta' attribute on this object to vary the steepness of the distribution (0 is a uniform distribution, increasing values lead to deterministic behavior). To use a different distribution altogether, simply override this method. |
Home | Trees | Indices | Help |
|
---|
Generated by Epydoc 3.0.1 on Wed Aug 19 16:47:44 2009 | http://epydoc.sourceforge.net |