Method applyReinforcementToLastAction

src/QLearn.cpp:78–87

}

void QLearn::applyReinforcementToLastAction(double reward, State newState) {
	if(lastState.size() == 0) {
		std::cout << "Called applyReinforcementToLastAction before an action had been selected! Because of this, this function call will be ignored.";
	}

	double lr = lastModel.network->getOutput(lastState)[0];
	double targetValueForLastState = lr + learningRate*(reward+(devaluationFactor*getHighestReward(newState))-lr);

	trainer->train(lastModel.network, {lastState}, {{targetValueForLastState}});
}

void QLearn::reset() {
	std::for_each(models.begin(), models.end(), [&](Model model) {