We study gradient descent under linearly correlated noise.

To reduce the computation burden, regularizing the OT problem with e.g. an entropic term

We consider the problem of minimizing a continuous function given given access to a natural quantum generalization of a stochastic gradient oracle. We provide two new methods for the special case of minimizing a Lipschitz convex function. Each method obtains a dimension versus accuracy trade-off which is provably unachievable classically and we prove that one method is asymptotically optimal in low-dimensional settings. Additionally, we provide quantum algorithms for computing a critical point of a smooth non-convex function at rates not known to be achievable classically. To obtain these results we build upon the quantum multivariate mean estimation result of Cornelissen et al. [25] and provide a general quantum variance reduction technique of independent interest.

In Appendix A.8, we examine the relationship between the training and test distributions In Appendix A.9, we present additional and extended experimental results that compare the In Appendix A.10, we provide additional experimental results that illustrate how the training In Appendix A.11, we include decision boundary visualizations for imbalanced training. In Appendix C, we discuss the limitations of our study. Lastly, in Appendix D, we discuss the broader impact of our work. This normalization allows us to examine the relative effect of batch size. Figure 7: Optimal augmentations depend on the imbalance ratio.

In reality, some events are common while others are exceedingly rare. As a result, machine learning systems, often developed in class-balanced settings [e.g.,

We study benign overfitting in two-layer ReLU networks trained using gradient descent and hinge loss on noisy data for binary classification.