Distributionally Robust Optimization with Probabilistic Group
نویسندگان
چکیده
Modern machine learning models may be susceptible to spurious correlations that hold on average but not for the atypical group of samples. To address problem, previous approaches minimize empirical worst-group risk. Despite promise, they often assume each sample belongs one and only group, which does allow expressing uncertainty in labeling. In this paper, we propose a novel framework PG-DRO, explores idea probabilistic membership distributionally robust optimization. Key our framework, consider soft instead hard annotations. The probabilities can flexibly generated using either supervised or zero-shot approaches. Our accommodates samples with ambiguity, offering stronger flexibility generality than prior art. We comprehensively evaluate PG-DRO both image classification natural language processing benchmarks, establishing superior performance.
منابع مشابه
Distributionally Robust Convex Optimization
Distributionally robust optimization is a paradigm for decision-making under uncertaintywhere the uncertain problem data is governed by a probability distribution that is itself subjectto uncertainty. The distribution is then assumed to belong to an ambiguity set comprising alldistributions that are compatible with the decision maker’s prior information. In this paper,we propose...
متن کاملDistributionally Robust Stochastic Optimization with Dependence Structure
Distributionally robust stochastic optimization (DRSO) is a framework for decision-making problems under certainty, which finds solutions that perform well for a chosen set of probability distributions. Many different approaches for specifying a set of distributions have been proposed. The choice matters, because it affects the results, and the relative performance of different choices depend o...
متن کاملDistributionally Robust Stochastic Optimization with Wasserstein Distance
Distributionally robust stochastic optimization (DRSO) is a robust approach to stochastic optimization problems in which the underlying distribution is not known exactly. It seeks a decision which hedges against the worst-case distribution in an ambiguity set, containing a family of distributions relevant to the considered problem. Unfortunately, the worst-case distributions resulting from many...
متن کاملTractable Distributionally Robust Optimization with Data
We present a unified and tractable framework for distributionally robust optimization that could encompass a variety of statistical information including, among others things, constraints on expectation, conditional expectation, and disjoint confidence sets with uncertain probabilities defined by φ-divergence. In particular, we also show that the Wasserstein-based ambiguity set has an equivalen...
متن کاملCalibration of Distributionally Robust Empirical Optimization Models
JUN-YA GOTOH, MICHAEL JONG KIM, AND ANDREW E.B. LIM Department of Industrial and Systems Engineering, Chuo University, Tokyo, Japan. Email: [email protected] Sauder School of Business, University of British Columbia, Vancouver, Canada. Email: [email protected] Departments of Decision Sciences and Finance, NUS Business School, National University of Singapore, Singapore. Email: andr...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the ... AAAI Conference on Artificial Intelligence
سال: 2023
ISSN: ['2159-5399', '2374-3468']
DOI: https://doi.org/10.1609/aaai.v37i10.26394