Abstract:We present an optimization-based framework to construct confidence intervals for functionals in constrained inverse problems, ensuring valid one-at-a-time frequentist coverage guarantees. Our approach builds upon the now-called strict bounds intervals, originally pioneered by Burrus (1965) and Rust and Burrus (1972), which offer ways to directly incorporate any side information about the parameters during inference without introducing external biases. This family of methods allows for uncertainty quantification in ill-posed inverse problems without needing to select a regularizing prior. By tying optimization-based intervals to an inversion of a constrained likelihood ratio test, we translate interval coverage guarantees into type I error control and characterize the resulting interval via solutions to optimization problems. Along the way, we refute the Burrus conjecture, which posited that, for possibly rank-deficient linear Gaussian models with positivity constraints, a correction based on the quantile of the chi-squared distribution with one degree of freedom suffices to shorten intervals while maintaining frequentist coverage guarantees. Our framework provides a novel approach to analyzing the conjecture, and we construct a counterexample employing a stochastic dominance argument, which we also use to disprove a general form of the conjecture. We illustrate our framework with several numerical examples and provide directions for extensions beyond the Rust-Burrus method for nonlinear, non-Gaussian settings with general constraints.

A Toolbox for Fast Interval Arithmetic in numpy with an Application to Formal Verification of Neural Network Controlled Systems

Forward Invariance in Neural Network Controlled Systems

Interval Reachability of Nonlinear Dynamical Systems with Neural Network Controllers

SAVER: A Toolbox for Sampling-Based, Probabilistic Verification of Neural Networks

$\texttt{immrax}$: A Parallelizable and Differentiable Toolbox for Interval Analysis and Mixed Monotone Reachability in JAX

The Fundamental Limits of Interval Arithmetic for Neural Networks

Bound Tightening using Rolling-Horizon Decomposition for Neural Network Verification

Efficient Interaction-Aware Interval Analysis of Neural Network Feedback Loops

ModelVerification.jl: a Comprehensive Toolbox for Formally Verifying Deep Neural Networks

Toward Neural-Network-Guided Program Synthesis and Verification

A Toolbox for Refined Information-Theoretic Analyses with Applications

Compositional Inductive Invariant Based Verification of Neural Network Controlled Systems

Verification of Neural Network Control Systems in Continuous Time

Computer-Assisted Verification of Four Interval Arithmetic Operators

Interval analysis for neural networks with application to fault detection

NNV: The Neural Network Verification Tool for Deep Neural Networks and Learning-Enabled Cyber-Physical Systems

Designing and developing Intuitionistic Fuzzy Logic Toolbox in MATLAB: Membership and non-membership functions gallery

Optimization-based frequentist confidence intervals for functionals in constrained inverse problems: Resolving the Burrus conjecture

Verifying Neural Network Controlled Systems Using Neural Networks

NeuroCodeBench: a plain C neural network benchmark for software verification

A toolbox of Equation-Free functions in Matlab\Octave for efficient system level simulation