Distribution-Free Statistical Dispersion Control for Societal Applications
Explicit finite-sample statistical guarantees on model performance are an important ingredient in responsible machine learning. Previous work has focused mainly on bounding either the expected loss of a predictor or the probability that an individual prediction will incur a loss value in a specified range. However, for many high-stakes applications, it is crucial to understand and control the dispersion of a loss distribution, or the extent to which different members of a population experience unequal effects of algorithmic decisions. We initiate the study of distribution-free control of statistical dispersion measures with societal implications and propose a simple yet flexible framework that allows us to handle a much richer class of statistical functionals beyond previous work. Our methods are verified through experiments in toxic comment detection, medical imaging, and film recommendation.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
How Much is Enough? An Empirical Test of the Resource Dispersion Hypothesis
Free-ranging dogs (Canis familiaris) thrive in diverse landscapes, including those heavily modified by humans. This study investigated the influence of resource availability on their spatial ecology across 52 rural and 4…
ManagementAsset Price Distributions and Efficient Markets
We explore a decomposition in which returns on a large class of portfolios relative to the market depend on a smooth non-negative drift and changes in the asset price distribution. This decomposition is obtained using ge…
AGPO: Adaptive Group Policy Optimization with Dual Statistical Feedback
Reinforcement learning improves LLM reasoning, but PPO/GRPO typically use fixed clipping and decoding temperature, which makes training brittle and tuning-heavy. We propose Adaptive Group Policy Optimization (AGPO), a cr…
Reinforcement LearningControlling Peak Sharpness in Multimodal Biomolecular Systems via the Chemical Fokker-Planck Equation
Intracellular biomolecular systems exhibit intrinsic stochasticity due to low molecular copy numbers, leading to multimodal probability distributions that play a crucial role in probabilistic differentiation and cellular…
Decision MakingTract Orientation and Angular Dispersion Deviation Indicator (TOADDI): A framework for single-subject analysis in diffusion tensor imaging
The purpose of this work is to develop a framework for single-subject analysis of diffusion tensor imaging (DTI) data. This framework (termed TOADDI) is capable of testing whether an individual tract as represented by th…