A Non-robust Model Training
–Neural Information Processing Systems
The model confidence distributions are shown in Figure 10 and Figure 11. Each row contains the same model adversarially and standard trained. In contrast, the robust models are better calibrated. Each row contains the same model adversarially and standard trained. In contrast, the robust models are better calibrated.
Neural Information Processing Systems
Nov-17-2025, 23:48:56 GMT
- Technology: