Training Deep Networks from Zero to Hero: avoiding pitfalls and going beyond

Abstract

Training deep neural networks may be challenging in real world data. Using models as black-boxes, even with transfer learning, can result in poor generalization or inconclusive results when it comes to small datasets or specific applications. This tutorial covers the basic steps as well as more recent options to improve models, in particular, but not restricted to, supervised learning. It can be particularly useful in datasets that are not as well-prepared as those in challenges, and also under scarce annotation and/or small data. We describe basic procedures: as data preparation, optimization and transfer learning, but also recent architectural choices such as use of transformer modules, alternative convolutional layers, activation functions, wide and deep networks, as well as training procedures including as curriculum, contrastive and self-supervised learning.

Optimization Matters: Guidelines to Improve Representation Learning with Deep Networks

“…Also, even for small networks, training is still known to be NP-hard [Bottou et al 2018]. Therefore it is paramount to make educated choices of optimization methods and training strategies [Ponti et al 2021]. Many studies focus on finding the best network architectures, while neglecting the optimization details, with unjustified or arbitrary choices, e.g.…”