Learning to See by Looking at Noise

Abstract

Current vision systems are trained on huge datasets, and these datasets come with costs: curation is expensive, they inherit human biases, and there are concerns over privacy and usage rights. To counter these costs, interest has surged in learning from cheaper data sources, such as unlabeled images. In this paper we go a step further and ask if we can do away with real image datasets entirely, instead learning from noise processes. We investigate a suite of image generation models that produce images from simple random processes. These are then used as training data for a visual representation learner with a contrastive loss. We study two types of noise processes, statistical image models and deep generative models under different random initializations. Our findings show that it is important for the noise to capture certain structural properties of real data but that good performance can be achieved even with processes that are far from realistic. We also find that diversity is a key property to learn good representations. Datasets, models, and code are available at https://mbaradad.github.io/learning_with_noise.

References in corpus (9)

• Learning Transferable Visual Models From Natural Language Supervision
• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
• What Makes for Good Views for Contrastive Learning?
• What Makes Good Synthetic Training Data for Learning Disparity and Optical Flow Estimation?
• A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark
• Self-supervised Pretraining of Visual Features in the Wild
• A critical analysis of self-supervision, or what we can learn from a single image
• InGAN: Capturing and Remapping the "DNA" of a Natural Image
• Multiscaling and information content of natural color images