Statistical Learning
StEik: Stabilizing the Optimization of Neural Signed Distance Functions and Finer Shape Representation Huizong Yang 1* Y uxin Sun
We show analytically how other terms added to the loss, currently used in the literature for other purposes, can actually eliminate these instabilities.
A Theory of Transfer-Based Black-Box Attacks: Explanation and Implications
This paper studies transfer-based attacks under a unified theoretical framework.
Improving the Privacy and Practicality of Objective Perturbation for Differentially Private Linear Learners
How does DP-SGD fare on simpler models such as linear and logistic regression?
A Unified Solution for Privacy and Communication Efficiency in Vertical Federated Learning (Supplemental Material) Anonymous Author(s) Affiliation Address email A Notation Table 1
Our achieved test accuracy is comparable to the SOT A VFL methodology.
Fast Conditional Mixing of MCMC Algorithms for Non-log-concave Distributions
This fast mixing guarantee can hold in cases when global mixing is provably slow.