Robust Imitation Learning With Weak Demonstrations