Sensitivity Analysis

Follow this topic Ask about this topic

Sensitivity analysis asks which inputs, assumptions, or model components drive an output, and how strongly their effects change across conditions. Common approaches include controlled perturbations, derivative and adjoint methods, variance-based measures such as Sobol indices, Shapley effects, and surrogate models for costly simulations. These tools are used to understand physical and agent-based models, assess causal conclusions, prioritize features, and audit machine-learning reliability and robustness. A central challenge is making analysis scalable while distinguishing sensitivity from causality or correctness: accurate predictions and strong average scores do not guarantee credible responses to specific changes.

Latest papers

39 papers6 papers in the last four weeks, up 20% on the four weeks before. 0.0% of all new papers.

Separating Geometry from Probability in the Analysis of Generalization
Apr 21, 2026Maxim Raginsky, Benjamin RechtStatistical Learning Theory Sensitivity Analysis
*University of Illinois, Urbana, IL 61801, USA · †University of California, Berkeley, CA 94720, USA
Analytical Extraction of Conditional Sobol' Indices via Basis Decomposition of Polynomial Chaos Expansions
Apr 21, 2026Shijie Zhong, Jiangfeng FuUncertainty Quantification Sensitivity Analysis
School of Power and Energy Northwestern Polytechnical University Xi’an, Shanxi 710129
Intrinsic perturbation scale for certified oracle objectives with epigraphic information
Apr 7, 2026Karim Bounja, Boujemaâ Achchab, Abdeljalil SakatSensitivity Analysis
Laboratory for Analysis and Modeling of Systems and Decision Support (LAMSAD), Hassan 1st University of Settat, Settat 26000, Morocco
From Model-Based Screening to Data-Driven Surrogates: A Multi-Stage Workflow for Exploring Stochastic Agent-Based Models
Apr 3, 2026Paul Saves, Matthieu Mastio, Nicolas Verstaevel +1Sensitivity Analysis Agent-Based Modeling
IRIT, Université Toulouse Capitole, Toulouse, France
Joint Surrogate Learning of Objectives, Constraints, and Sensitivities for Efficient Multi-objective Optimization of Neural Dynamical Systems
Mar 22, 2026Frithjof Gressmann, Ivan Georgiev Raikov, Seung Hyun Kim +3Neural Surrogate Modeling Surrogate-Assisted Optimization
Siebel School of Computing and Data Science, University of Illinois Urbana-Champaign, Urbana, IL · The Grainer College of Engineering, University of Illinois Urbana-Champaign, Urbana, IL · Department of Neurosurgery, Stanford University, Stanford, CA +3
How Wrong Can Your Counterfactual Be? Quantifying Confounding Bias for Continuous Treatments without a Control Group
Mar 8, 2026Yu Wang, Xiangchen Liu, Siguang LiCausal Effect Estimation Causal Inference
Department of Economics, Cornell University. · Department of Family and Consumer Sciences, California State University, Long Beach. · Society Hub, Hong Kong University of Science and Technology (Guangzhou).
Exact Attention Sensitivity and the Geometry of Transformer Stability
Feb 21, 2026Seyed Morteza EmadiTransformer Attention Softmax Attention
UNC-Chapel Hill, Kenan-Flagler Business School Chapel Hill, NC, USA
White-Box Sensitivity Auditing with Steering Vectors
Jan 23, 2026Hannah Cyberey, Yangfeng Ji, David EvansAlgorithmic Auditing LLM Auditing
University of Virginia
Uncertainty measurement for complex event prediction in safety-critical systems
Nov 2, 2024Maria J. P. Peixoto, Akramul AzimUncertainty Quantification Sensitivity Analysis
Department of Electrical, Computer and Software Engineering, Ontario Tech University, Oshawa, Ontario, Canada.

Apr 21, 2026J/K move · Enter open · S save

Separating Geometry from Probability in the Analysis of Generalization

Maxim Raginsky, Benjamin Recht

*University of Illinois, Urbana, IL 61801, USA · †University of California, Berkeley, CA 94720, USA

Statistical Learning Theory Sensitivity Analysis Generalization Bounds

Open paper AskPDF arXiv

The goal of machine learning is to find models that minimize prediction error on data that has not yet been seen. Its operational paradigm assumes access to a dataset

S

and articulates a scheme for evaluating how well a given model performs on an arbitrary sample. The sample can be

S

(in which case we speak of in-sample'' performance) or some entirely new $S'$ (in which case we speak of out-of-sample'' performance). Traditional analysis of generalization assumes that both in- and out-of-sample data are i.i.d.\ draws from an infinite population. However, these probabilistic assumptions cannot be verified even in principle. This paper presents an alternative view of generalization through the lens of sensitivity analysis of solutions of optimization problems to perturbations in the problem data. Under this framework, generalization bounds are obtained by purely deterministic means and take the form of variational principles that relate in-sample and out-of-sample evaluations through an error term that quantifies how close out-of-sample data are to in-sample data. Statistical assumptions can then be used ex post to characterize the situations when this error term is small (either on average or with high probability).