[INTERSPEECH 2020] Defense for Black-box Attacks on Anti-spoofing Models by Self-Supervised Learning | Hung-yi Lee | Podwise