Timezone: »

Characterizing Generalization under Out-Of-Distribution Shifts in Deep Metric Learning
Timo Milbich · Karsten Roth · Samarth Sinha · Ludwig Schmidt · Marzyeh Ghassemi · Bjorn Ommer

Thu Dec 09 08:30 AM -- 10:00 AM (PST) @ None #None

Deep Metric Learning (DML) aims to find representations suitable for zero-shot transfer to a priori unknown test distributions. However, common evaluation protocols only test a single, fixed data split in which train and test classes are assigned randomly. More realistic evaluations should consider a broad spectrum of distribution shifts with potentially varying degree and difficulty.In this work, we systematically construct train-test splits of increasing difficulty and present the ooDML benchmark to characterize generalization under out-of-distribution shifts in DML. ooDML is designed to probe the generalization performance on much more challenging, diverse train-to-test distribution shifts. Based on our new benchmark, we conduct a thorough empirical analysis of state-of-the-art DML methods. We find that while generalization tends to consistently degrade with difficulty, some methods are better at retaining performance as the distribution shift increases. Finally, we propose few-shot DML as an efficient way to consistently improve generalization in response to unknown test shifts presented in ooDML.

Author Information

Timo Milbich (LMU Munich & Heidelberg University)
Karsten Roth (University of Tuebingen)
Samarth Sinha (University of Toronto, Vector Institute)
Ludwig Schmidt (University of Washington)
Marzyeh Ghassemi (University of Toronto, Vector Institute)
Bjorn Ommer (Heidelberg University)

More from the Same Authors