Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Affiliation:
3
0
3
6
Merging differentially private models can be dramatically improved by addressing geometric obstacles, leading to better performance without sacrificing privacy.
Noise-aware shrinkage can enhance the utility of differentially private fine-tuning, outperforming traditional methods without compromising privacy or efficiency.
Geometry-consistent aggregation can boost federated learning accuracy by over 2% while maintaining differential privacy.