An Abstraction-Refinement Approach to Formal Verification of Tree Ensembles

Törnblom, John; Nadjm‐Tehrani, Simin

doi:10.1007/978-3-030-26250-1_24

Cited by 19 publications

(23 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Related Work. It is only recently that adversarial attacks and robustness of tree ensemble classifiers started to be a subject of investigation (Andriushchenko and Hein 2019;Chen et al 2019a;Einziger et al 2019;Kantchelian, Tygar, and Joseph 2016;Törnblom and Nadjm-Tehrani 2018;Törnblom and Nadjm-Tehrani 2019b;Törnblom and Nadjm-Tehrani 2019a). The most related work is the robustness verification tool of tree ensembles by Törnblom and Nadjm-Tehrani (2019b) called VoTE.…”

Section: Introductionmentioning

confidence: 99%

Abstract Interpretation of Decision Tree Ensemble Classifiers

Ranzato

Zanella

2020

AAAI

View full text Add to dashboard Cite

We study the problem of formally and automatically verifying robustness properties of decision tree ensemble classifiers such as random forests and gradient boosted decision tree models. A recent stream of works showed how abstract interpretation, which is ubiquitously used in static program analysis, can be successfully deployed to formally verify (deep) neural networks. In this work we push forward this line of research by designing a general and principled abstract interpretation-based framework for the formal verification of robustness and stability properties of decision tree ensemble models. Our abstract interpretation-based method may induce complete robustness checks of standard adversarial perturbations and output concrete adversarial attacks. We implemented our abstract verification technique in a tool called silva, which leverages an abstract domain of not necessarily closed real hyperrectangles and is instantiated to verify random forests and gradient boosted decision trees. Our experimental evaluation on the MNIST dataset shows that silva provides a precise and efficient tool which advances the current state of the art in tree ensembles verification.

show abstract

Section: Introductionmentioning

confidence: 99%

Abstract Interpretation of Decision Tree Ensemble Classifiers

Ranzato

Zanella

2020

AAAI

View full text Add to dashboard Cite

show abstract

“…Moreover, computing all the equivalence classes of an ensemble is infeasible in general due to their combinatorial explosion, as also observed by the authors of VoTE. In a follow-up work, the same authors proposed an abstraction-refinement approach to mitigate this complexity problem [19]. However, contrary to our analysis, their extension is not proved sound, which is an important requirement for the analysis of classifiers deployed in adversarial settings.…”

Section: B Security Of Decision Treesmentioning

confidence: 75%

Beyond Robustness: Resilience Verification of Tree-Based Classifiers

Calzavara¹,

Lorenzo²,

Lucchese³

et al. 2021

Preprint

View full text Add to dashboard Cite

In this paper we criticize the robustness measure traditionally employed to assess the performance of machine learning models deployed in adversarial settings. To mitigate the limitations of robustness, we introduce a new measure called resilience and we focus on its verification. In particular, we discuss how resilience can be verified by combining a traditional robustness verification technique with a data-independent stability analysis, which identifies a subset of the feature space where the model does not change its predictions despite adversarial manipulations. We then introduce a formally sound data-independent stability analysis for decision trees and decision tree ensembles, which we experimentally assess on public datasets and we leverage for resilience verification. Our results show that resilience verification is useful and feasible in practice, yielding a more reliable security assessment of both standard and robust decision tree models.

show abstract

“…Using insights from this comparison, we then show that parallelisation of VoTE computations is possible with little additional effort. -We illustrate that VoTE is capable of verifying properties different from robustness (treated earlier [19]) through the modularity of the property checker. This is done by replicating the verification of ten properties of an aircraft collision avoidance system previously studied in the context of neural networks [11].…”

Section: Introductionmentioning

confidence: 98%

“…This paper studies formal verification of tree ensembles, a class of ML models considered less opaque than other ML models, thus possessing more intrinsic explainability. The paper is centred around the tool VoTE, that uses abstract interpretation to reduce the search space when verifying a given property, and was presented informally earlier [19]. We aim to show that, in addition to building on well-known mathematical concepts (conservative approximations and abstraction-refinement) that make the verification algorithm sound and complete, the tool set can be extended towards useful practical properties that are a result of architectural and algorithmic design decisions.…”

Section: Introductionmentioning

confidence: 99%

Scaling up Memory-Efficient Formal Verification Tools for Tree Ensembles

Törnblom¹,

Nadjm‐Tehrani²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

To guarantee that machine learning models yield outputs that are not only accurate, but also robust, recent works propose formally verifying robustness properties of machine learning models. To be applicable to realistic safety-critical systems, the used verification algorithms need to manage the combinatorial explosion resulting from vast variations in the input domain, and be able to verify correctness properties derived from versatile and domain-specific requirements. In this paper, we formalise the VoTE algorithm presented earlier as a tool description, and extend the tool set with mechanisms for systematic scalability studies. In particular, we show a) how the separation of property checking from the core verification engine enables verification of versatile requirements, b) the scalability of the tool, both in terms of time taken for verification and use of memory, and c) that the algorithm has attractive properties that lend themselves well for massive parallelisation. We demonstrate the application of the tool in two case studies, namely digit recognition and aircraft collision avoidance, where the first case study serves to assess the resource utilisation of the tool, and the second to assess the ability to verify versatile correctness properties.

show abstract

An Abstraction-Refinement Approach to Formal Verification of Tree Ensembles

Cited by 19 publications

References 9 publications

Abstract Interpretation of Decision Tree Ensemble Classifiers

Abstract Interpretation of Decision Tree Ensemble Classifiers

Beyond Robustness: Resilience Verification of Tree-Based Classifiers

Scaling up Memory-Efficient Formal Verification Tools for Tree Ensembles

Contact Info

Product

Resources

About