On Pre-Training for Visuo-Motor Control: Revisiting a Learning-from-Scratch Baseline

Nicklas Hansen, Zhecheng Yuan, Yanjie Ze, Tongzhou Mu, Aravind Rajeswaran, Hao Su, Huazhe Xu, Xiaolong Wang

Keywords: to pre-train, not to pre-train

TL;DR: We revisit a simple Learning-from-Scratch baseline for visuo-motor control that uses data augmentation and a shallow ConvNet, and show that it is competitive with recent methods that leverage frozen pre-trained visual representations.

Abstract: We revisit a simple Learning-from-Scratch baseline for visuo-motor control that uses data augmentation and a shallow ConvNet. We find that this baseline has competitive performance with recent methods that leverage frozen visual representations trained on large-scale vision datasets.