Search papers, labs, and topics across Lattice.
University of California, San Diego
1
0
2
Alignment-based defenses can effectively shield LLMs from property inference attacks without requiring access to original training data.
