We prove that early learning and memorization are fundamental phenomena in high-dimensional classification tasks, even in simple linear models, and give a theoretical explanation in this setting.

The selection of the step-size is of great importance in both the theoretical and practical performance of first-order methods, including the aforementioned VR methods.

For a full version of this paper see [53].

Self-supervised multi-view stereopsis (MVS) attracts increasing attention for learning dense surface predictions from only a set of images without onerous ground-truth 3D training data for supervision.