International audienceWe present a new parameterized abstract domain able to refine existing numerical abstract domains with finite disjunctions. The elements of the abstract domain are decision trees where the decision nodes are labeled with linear constraints, and the leaf nodes belong to a numerical abstract domain. The abstract domain is parametric in the choice between the expressivity and the cost of the linear constraints for the decision nodes (e.g., polyhedral or octagonal constraints), and the choice of the abstract domain for the leaf nodes. We describe an instance of this domain based on piecewise-defined ranking functions for the automatic inference of sufficient preconditions for program termination. We have implemented a static analyzer for proving conditional termination of programs written in (a subset of) C and, using experimental evidence, we show that it performs well on a wide variety of benchmarks, it is competitive with the state of the art and is able to analyze programs that are out of the reach of existing methods
International audienceWe present a parameterized abstract domain for proving program termination by abstract interpretation. The domain automatically synthesizes piecewise-defined ranking functions and infers sufficient conditions for program termination. The analysis uses over-approximations but we prove its soundness, meaning that all program executions respecting these sufficient conditions are indeed terminating. The abstract domain is parameterized by a numerical abstract domain for environments and a numerical abstract domain for functions. This parameterization allows to easily tune the trade-off between precision and cost of the analysis. We describe an instantiation of this generic do- main with intervals and affine functions. We define all abstract operators, including widening to ensure convergence. To illustrate the potential of the proposed framework, we have implemented a research prototype static analyzer, for a small imperative language, that yielded interesting preliminary results
Recently, there is growing concern that machine-learned software, which currently assists or even automates decision making, reproduces, and in the worst case reinforces, bias present in the training data. The development of tools and techniques for certifying fairness of this software or describing its biases is, therefore, critical. In this paper, we propose a perfectly parallel static analysis for certifying fairness of feed-forward neural networks used for classification of tabular data. When certification succeeds, our approach provides definite guarantees, otherwise, it describes and quantifies the biased input space regions. We design the analysis to be sound, in practice also exact, and configurable in terms of scalability and precision, thereby enabling pay-as-you-go certification. We implement our approach in an open-source tool called libra and demonstrate its effectiveness on neural networks trained on popular datasets. CCS Concepts: • Software and its engineering → Formal software verification; • Theory of computation → Program analysis; Abstraction; • Computing methodologies → Neural networks; • Social and professional topics → Computing / technology policy.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.