- Keywords: invariance, representation learning, out-of-distribution, domain adaptation, generalization
- Abstract: We introduce a general approach, called invariance through inference, for improving the test-time performance of a behavior agent in deployment environments with unknown perceptual variations. Instead of producing invariant visual features through memorization, invariance through inference turns adaptation at deployment-time into an unsupervised learning problem by trying to match the distribution of latent features to the agent's prior experience without relying on paired data. Although simple, we show that this idea leads to surprising improvements on a variety of adaptation scenarios without access to task reward, including changes in camera poses from the challenging distractor control suite.
- One-sentence Summary: Enable out-of-distribution generalization via invariance through inference
- Supplementary Material: zip