self.state_mean = self.state_mean * self.alpha + \
            observation.mean() * (1 - self.alpha)