no code implementations • 7 Dec 2018 • Zhongshu Gu, Hani Jamjoom, Dong Su, Heqing Huang, Jialong Zhang, Tengfei Ma, Dimitrios Pendarakis, Ian Molloy
We also demonstrate that when malicious training participants tend to implant backdoors during model training, CALTRAIN can accurately and precisely discover the poisoned and mislabeled training data that lead to the runtime mispredictions.