Gated Domain Units for Multi-source Domain Generalization
Abstract: The phenomenon of distribution shift (DS) occurs when a dataset at test time differs from the dataset at training time, which can significantly impair the performance of a machine learning model in practical settings due to a lack of knowledge about the data's distribution at test time. To address this problem, we postulate that real-world distributions are composed of latent Invariant Elementary Distributions (I.E.D) across different domains. This assumption implies an invariant structure in the solution space that enables knowledge transfer to unseen domains. To exploit this property for domain generalization, we introduce a modular neural network layer consisting of Gated Domain Units (GDUs) that learn a representation for each latent elementary distribution. During inference, a weighted ensemble of learning machines can be created by comparing new observations with the representations of each elementary distribution. Our flexible framework also accommodates scenarios where explicit domain information is not present. Extensive experiments on image, text, and graph data show consistent performance improvement on out-of-training target domains. These findings support the practicality of the I.E.D assumption and the effectiveness of GDUs for domain generalisation.
License: Creative Commons Attribution 4.0 International (CC BY 4.0)
Submission Length: Long submission (more than 12 pages of main content)
Changes Since Last Submission: We discussed simulating multi-domain data using Gaussian mixtures for the simulation experiments, and investigating whether the GDUs can approximate the components. However, this experiment is quite similar to the digits data experiments, and the contribution is not significant enough to justify a separate section in the paper, including another round of discussion. Hence, we decided to publish the paper without additional simulation experiments.
Supplementary Material: zip
Assigned Action Editor: ~Nicolas_THOME2
Submission Number: 1168