parameterName = value
The list of current parameters is in the following table. The sample values given correspond to a set-up where active exploration is used and MACS performs 10 random planning steps per time step. You can generally use one of the MACS parameters file "as such".
The active exploration parameters are stored in a list. The user must first give the number of criteria, and then, for each criterion, give its name (among "info","rehearsal","stateSearch" and "external", so far), followed by its rank number, and some parameters going with these criteria. The order is important: criterion number one will have priority over criterion number two, and so on.
parameter name | sample value | meaning |
systemName | MACS | Could be YACS or any other... |
General learning parameters | ||
initialReliability | 0.5 | The value of classifier reliability when created |
learningRate | 0.5 | The famous beta in Widrow-Hoff |
eGreedy | 0.0 | epsilon value for exploration |
nbPlanningSteps | 10 | number of value iteration steps on the model as in Dyna |
plannerType | RandomPlanner or SystematicPlanner | if random, nbPlanningSteps chosen randomly ; if systematic, nbPlanningSteps iterations on all known states |
classifierMemorySize | 5 | number of good and bad markers recorded |
nbEvaluationsToSpecialize | 5 | |
nbFailuresToRemove | 5 | |
nbSuccessToProbe | 5 | |
Active exploration parameters | Used for active exploration. See above | |
nbCriteria | 3 | |
criteriaNames0 | info | the information reward |
criteriaDiscountFactor0 | 0.5 | |
criteriaDefaultValue0 | 1.0 | |
criteriaIncrLearningRate0 | 0.1 | |
criteriaDecrLearningRate0 | 1.0 | |
criteriaNames1 | external | the external reward |
criteriaDiscountFactor1 | 0.5 | |
criteriaDefaultValue1 | 0.0 | |
criteriaIncrLearningRate1 | 1.0 | |
criteriaDecrLearningRate1 | 1.0 | |
criteriaNames2 | rehearsal | the rehearsal reward |
criteriaDiscountFactor2 | 0.5 | |
criteriaDefaultValue2 | 0.0 | |
criteriaIncrLearningRate2 | 0.1 | |
criteriaDecrLearningRate2 | 1.0 | |
General switches | ||
conditionCovering | on | |
obsoleteConditionSelection | on | ??? still used ? |
anticipationLearning | on | |
mutspecDiscovering | on | |
generalization | on | |
Trace parameters | ||
traceConditionCovering | on | Adds a message each time condition covering creates a classifier |
traceObsoleteConditionSelection | on | ??? still used ? |
traceAnticipationCovering | on | ??? still used ? |
traceAnticipationSelection | on | ??? still used ? |
traceMutspecCreation | on | Adds a message each time mutspec creates a classifier |
traceMutspecRemoval | on | Adds a message each time mutspec removes a classifier |
traceMutgenCreation | on | Adds a message each time generalization creates a classifier |
traceMutgenRemoval | on | Adds a message each time generalization removes a classifier |
traceStatistics | off | ? error bars ? |
traceGeneralInfos | on | nb classifiers, nb good, % knowledge... |
traceInputs | on | shows the input at each time step |
traceOutput | on | shows the action chosen at each time step |
traceClassifierListFrequency | 1000 | shows the complete classifier list at each interval |
traceSituationValuesFrequency | 1000 | shows the values of all known states at each interval |
traceTime | on | shows the number of time steps |