CLIP yielded impressive results on zero-shot transfer learning tasks and is considered as a
foundation model like BERT or GPT3. CLIP vision models that have a rich representation are …

Many black-box optimization tasks arising in high-stakes applications require risk-averse
decisions. The standard Bayesian optimization (BO) paradigm, however, optimizes the …

Differentially private learning algorithms inject noise into the learning process. While the
most common private learning algorithm, DP-SGD, adds independent Gaussian noise in …

Differentially private (stochastic) gradient descent is the workhorse of DP private machine
learning in both the convex and non-convex settings. Without privacy constraints, second …

We introduce a new PAC-Bayes oracle bound for unbounded losses. This result can be
understood as a PAC-Bayesian version of the Cram\'er-Chernoff bound. The proof technique …

We present a stochastic descent algorithm for unconstrained optimization that is particularly
efficient when the objective function is slow to evaluate and gradients are not easily …

We introduce a new PAC-Bayes oracle bound for unbounded losses that extends Cramér-
Chernoff bounds to the PAC-Bayesian setting. The proof technique relies on controlling the …

Consider a channel where is an-dimensional random vector, and is a multivariate Gaussian
vector with a full-rank covariance matrix. The object under consideration in this paper is the …

We explore probability distributions on the real line whose Laplace transform admits an
upper bound of subgaussian type known as strict subgaussianity. One class in this family …

This paper provides a general derivative identity for the conditional mean estimator of an
arbitrary vector signal in Gaussian noise with an arbitrary covariance matrix. This new …