Search papers, labs, and topics across Lattice.
University of Texas
3
0
7
Excluding features based on manipulability can lead to suboptimal predictions, revealing a critical flaw in standard feature selection practices.
DRA outputs are surprisingly variable, with inference and early-stage decisions being the biggest culprits, but structured outputs and ensemble querying can significantly reduce this stochasticity.
Self-distillation isn't just a trick: this paper proves it *provably* improves ridge regression performance, even with negative mixing weights in over-regularized regimes, and offers a one-shot tuning method to make it practical.
