public Transition(State startState, State endState, string symbolName, List <string> Delta)
{
this.startState = startState;
this.endState = endState;
this.symbolName = symbolName;
distribution = new OutputDistribution(Delta);
expecations = new Expectation();
}
public void ApplyPunishment()
{
float t = 1;
Transition marked = markedDistributions.Pop();
while (marked != null)
{
OutputDistribution distribution = marked.GetOutputDistribution();
foreach (string name in markedSymbols)
{
float change = parameters [3] * t *curSymbol.GetValue() / marked.GetConfidence();
distribution.UpdateSymbolProbabilityForAllBut(name, change);
marked.AddToConfidence(parameters [3] * t * curSymbol.GetValue());
foreach (State state in states)
{
Transition nextTransition = state.GetTransitionOn(new Symbol(marked.GetSymbolName(), 0.0f));
if (nextTransition != null)
{
nextTransition.GetOutputDistribution().UpdateSymbolProbabilityForAllBut(name, change);
nextTransition.AddToConfidence(parameters [6] * parameters [3] * t * curSymbol.GetValue());
}
}
}
t *= parameters [7];
if (markedDistributions.Count > 0)
{
marked = markedDistributions.Pop();
}
else
{
marked = null;
}
}
markedSymbols = new List <String> ();
}
private void BuildModel()
{
List <string> Sigma = new List <string>();
List <string> Delta = new List <string>();
List <State> states = new List <State>();
List <State> rewards = new List <State>();
List <State> punishments = new List <State>();
Sigma.AddRange(new string[] { "1N", "1S", "1W", "1E", "2N", "2S", "2W", "2E", "F" });
Delta.AddRange(new string[] { "LEFT", "RIGHT", "FORWARD" });
State s0 = new State("0"); //Default
State s1 = new State("1"); //1N
State s2 = new State("2"); //2N
State s3 = new State("3"); //1E
State s4 = new State("4"); //2E
State s5 = new State("5"); //1S
State s6 = new State("6"); //2S
State s7 = new State("7"); //1W
State s8 = new State("8"); //2W
State s9 = new State("9"); //F
states.AddRange(new State[] { s1, s2, s3, s4, s5, s6, s7, s8, s9 });
rewards.AddRange(new State[] { s9 });
punishments.AddRange(new State[] {});
Transition transition0_1 = new Transition(s0, s1, "1N", Delta);
Transition transition0_2 = new Transition(s0, s2, "2N", Delta);
Transition transition0_3 = new Transition(s0, s3, "1E", Delta);
Transition transition0_4 = new Transition(s0, s4, "2E", Delta);
Transition transition0_5 = new Transition(s0, s5, "1S", Delta);
Transition transition0_6 = new Transition(s0, s6, "2S", Delta);
Transition transition0_7 = new Transition(s0, s7, "1W", Delta);
Transition transition0_8 = new Transition(s0, s8, "2W", Delta);
Transition transition0_9 = new Transition(s0, s9, "F", Delta);
transition0_1.getExpectations().SetConfidence(10000);
transition0_2.getExpectations().SetConfidence(10000);
transition0_3.getExpectations().SetConfidence(10000);
transition0_4.getExpectations().SetConfidence(10000);
transition0_5.getExpectations().SetConfidence(10000);
transition0_6.getExpectations().SetConfidence(10000);
transition0_7.getExpectations().SetConfidence(10000);
transition0_8.getExpectations().SetConfidence(10000);
transition0_9.getExpectations().SetConfidence(10000);
s0.AddTransition(transition0_1);
s0.AddTransition(transition0_2);
s0.AddTransition(transition0_3);
s0.AddTransition(transition0_4);
s0.AddTransition(transition0_5);
s0.AddTransition(transition0_6);
s0.AddTransition(transition0_7);
s0.AddTransition(transition0_8);
s0.AddTransition(transition0_9);
OutputDistribution canForward = (new OutputDistribution(Delta));
canForward.SetDistribution(new List <Symbol>()
{
new Symbol("LEFT", 0.333f), new Symbol("RIGHT", 0.333f), new Symbol("FORWARD", 0.333f), new Symbol("", 0f)
});
OutputDistribution cannotForward = (new OutputDistribution(Delta));
cannotForward.SetDistribution(new List <Symbol> ()
{
new Symbol("LEFT", 0.5f), new Symbol("RIGHT", 0.5f), new Symbol("FORWARD", 0.0f), new Symbol("", 0f)
});
transition0_1.SetDistribution(cannotForward);
transition0_2.SetDistribution(cannotForward);
transition0_3.SetDistribution(canForward);
transition0_4.SetDistribution(canForward);
transition0_5.SetDistribution(cannotForward);
transition0_6.SetDistribution(cannotForward);
transition0_7.SetDistribution(cannotForward);
transition0_8.SetDistribution(canForward);
model = new Model(Sigma, Delta, states, s0, rewards, punishments);
}
public void SetDistribution(OutputDistribution distribution)
{
this.distribution = distribution;
}