Search papers, labs, and topics across Lattice.
2
0
3
Standard quadratic regularization fails to contain runaway client drift under extreme heterogeneity, but tuning the proximal exponent to $p \in [5, 7]$ slashes severe-stress federated loss by over 23%.
Recommender systems can achieve significantly better accuracy *and* diversity using hybrid multi-objective algorithms inspired by evolutionary optimization, outperforming traditional methods.
