Search papers, labs, and topics across Lattice.
2
0
4
0
Even if adversaries try to scrub the watermark, Cert-LAS offers provable guarantees for verifying ownership of text-to-image diffusion models.
Time series forecasting models are surprisingly susceptible to backdoor attacks due to data entanglement and task-formulation shift, but a new channel-wise pool training method called TimeGuard can effectively defend against them.
