Contrastively-trained Structured World Models (C-SWMs) depart from traditional pixel-based reconstruction losses and use an energy-based hinge loss for learning object-centric world models.
Analysis of invariances in representations from contrastive self-supervised models reveals that they leverage aggressive cropping on object-centric datasets to improve occlusion invariance at the expense of viewpoint and category instance invariance.
BYOL improves on SotA self-supervised methods by introducing a target network, which removes the need for negative examples.
SimCLR, a simple unsupervised contrastive learning framework, uses data augmentation for positive pairs, a nonlinear projection head, normalized temperature-scaled cross entropy loss, and large batch sizes to achieve SotA in self-supervised, semi-supervised, and transfer learning domains.
Development of artificial neural networks should leverage the insight that much of animal behavior is innate as a result of wiring rules encoded in the genome, learned through billions of years of evolution.
A large scale, comprehensive study challenges various assumptions in learning disentangled representations, which motivates demonstrating concrete benefits in robust experimental setups in future work.