Batch Normalization Embeddings for Deep Domain Generalization

Mattia Segù, Alessio Tonioni, Federico Tombari

Keywords: Domain Generalization, Domain Representation, Multi-source Domain Generalization

Abstract: Domain generalization aims at training machine learning models to perform robustly across different and unseen domains. Several recent methods use multiple datasets to train models to extract domain-invariant features, hoping to generalize to unseen domains. Instead, we explicitly train domain-dependant representations by using ad-hoc batch normalization layers to collect independent domain’s statistics. We propose to use these statistics to map domains in a shared latent space, where membership to a domain can be measured by means of a distance function. At test time, we project samples from an unknown domain into the same space and express their domain as a linear combination of the known ones. We apply the same mapping strategy at training and test time, learning both a latent representation and a powerful but lightweight ensemble model. We show a significant increase in classification accuracy over current state-of-the-art techniques on popular domain generalization benchmarks: PACS, Office-31 and Office-Caltech.

One-sentence Summary: Batch normalization representations allow to map domains in a latent space

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Community Implementations: [ 2 code implementations](https://www.catalyzex.com/paper/arxiv:2011.12672/code)

Reviewed Version (pdf): https://openreview.net/references/pdf?id=_iq88Q9A75