| Interface | Description |
|---|---|
| DenseStateActionFeatures | |
| DenseStateFeatures |
Many functions approximation techniques require a fixed feature vector to work and in many cases, using abstract features from
the state attributes is useful.
|
| Class | Description |
|---|---|
| ConcatenatedObjectFeatures |
This class is used to produce a state feature vector from an
OOState by iterating
over the objects, generating a double array for each object, and concatenating the reslting vectors into one vector. |
| DenseCrossProductFeatures |
Class that generates state-action features as cross product of underlying state-features with the action set.
|
| DenseLinearVFA |
This class can be used to perform linear value function approximation, either for a states or state-actions (Q-values).
|
| DenseStateActionLinearVFA | |
| NormalizedVariableFeatures |
This class is will construct a double array from states by iterating over numeric state variables and setting values
in the double array to the variables normalized value.
|
| NumericVariableFeatures |
A
DenseStateFeatures that iterates through all state variables in a state
and places them into the returned double array. |
| PFFeatures |
Binary features that are determined from a list of
PropositionalFunctions. |
| SparseToDenseFeatures |
A wrapper for turning the features from a
SparseStateFeatures into a DenseStateFeatures. |