Search papers, labs, and topics across Lattice.
This paper addresses the problem of set representation learning models being vulnerable to element-level corruptions at inference time. They introduce SW-DRSO, a distributionally robust optimization framework that optimizes for the worst-case expected loss over plausible inference-time variations using a barycentric adversary. Experiments on four tasks demonstrate that SW-DRSO improves robustness against corruption without sacrificing overall performance.
Set representation models can be made robust to inference-time corruptions like outliers and missing data by training against a learned barycentric adversary.
Standard Set Representation Learning methods typically excel on curated data but often overlook the challenge of inference-time element corruption. This refers to scenarios where deployed models encounter element-level degradations, such as outliers or missing components, that may distort set representation and degrade performance. We propose SW-DRSO, a distributionally robust optimization framework tailored for sets. Rather than minimizing loss solely on observed training data, SW-DRSO optimizes a tractable surrogate of the worst-case expected loss over a family of plausible inference-time variations. We introduce a barycentric adversary that approximates the intractable search over corrupted sets by a differentiable training-time optimization over simplex weights. Extensive experiments across four tasks demonstrate that SW-DRSO effectively enhances robustness against corruption while maintaining high overall performance.
