Improving upper bounds for the distinguishing index