Search papers, labs, and topics across Lattice.
Affiliation:
1
0
2
Sparse backdoor attacks can misclassify over half of triggered anomalies while keeping clean accuracy losses below 2%, revealing a stealthy threat in federated learning systems.
