The SSL Interplay: Augmentations, Inductive Bias, and Generalization

Vivien Cabannes, Bobak T. Kiani, Randall Balestriero, Yann LeCun, Alberto Bietti

Research output: Contribution to journalConference articlepeer-review


Self-supervised learning (SSL) has emerged as a powerful framework to learn representations from raw data without supervision. Yet in practice, engineers face issues such as instability in tuning optimizers and collapse of representations during training. Such challenges motivate the need for a theory to shed light on the complex interplay between the choice of data augmentation, network architecture, and training algorithm. We study such an interplay with a precise analysis of generalization performance on both pretraining and downstream tasks in a theory friendly setup, and highlight several insights for SSL practitioners that arise from our theory.

Original languageEnglish (US)
Pages (from-to)3252-3298
Number of pages47
JournalProceedings of Machine Learning Research
StatePublished - 2023
Event40th International Conference on Machine Learning, ICML 2023 - Honolulu, United States
Duration: Jul 23 2023Jul 29 2023

ASJC Scopus subject areas

  • Artificial Intelligence
  • Software
  • Control and Systems Engineering
  • Statistics and Probability


Dive into the research topics of 'The SSL Interplay: Augmentations, Inductive Bias, and Generalization'. Together they form a unique fingerprint.

Cite this