Search

Neural network verification is a new and rapidly developing field of research. So far, the main priority has been establishing efficient verification algorithms and tools, while proper support from the programming language perspective has been considered secondary or unimportant. Yet, there is mounting evidence that insights from the programming language community may make a difference in the future development of this domain. In this paper, we formulate neural network verification challenges as programming language challenges and suggest possible future solutions., Comment: Accepted at ESOP 2025, European Symposium on Programming Languages

Training large language models (LLMs) requires a substantial investment of time and money. To get a good return on investment, the developers spend considerable effort ensuring that the model never produces harmful and offensive outputs. However, bad-faith actors may still try to slander the reputation of an LLM by publicly reporting a forged output. In this paper, we show that defending against such slander attacks requires reconstructing the input of the forged output or proving that it does not exist. To do so, we propose and evaluate a search based approach for targeted adversarial attacks for LLMs. Our experiments show that we are rarely able to reconstruct the exact input of an arbitrary output, thus demonstrating that LLMs are still vulnerable to slander attacks., Comment: 4 pages, 3 figures

Bounded Model Checking (BMC) is a widely used software verification technique. Despite its successes, the technique has several limiting factors, from state-space explosion to lack of completeness. Over the years, interval analysis has repeatedly been proposed as a partial solution to these limitations. In this work, we evaluate whether the computational cost of interval analysis yields significant enough improvements in BMC's performance to justify its use. In more detail, we quantify the benefits of interval analysis on two benchmarks: the Intel Core Power Management firmware and 9537 programs in the ReachSafety category of the International Competition on Software Verification. Our results show that interval analysis is essential in solving 203 unique benchmarks., Comment: Submitted to IFM

Realm Management Monitor (RMM) is an essential firmware component within the recent Arm Confidential Computing Architecture (Arm CCA). Previous work applies formal techniques to verify the specification and prototype reference implementation of RMM. However, relying solely on a single verification tool may lead to the oversight of certain bugs or vulnerabilities. This paper discusses the application of ESBMC, a state-of-the-art Satisfiability Modulo Theories (SMT)-based software model checker, to further enhance RRM verification. We demonstrate ESBMC's ability to precisely parse the source code and identify specification failures within a reasonable time frame. Moreover, we propose potential improvements for ESBMC to enhance its efficiency for industry engineers. This work contributes to exploring the capabilities of formal verification techniques in real-world scenarios and suggests avenues for further improvements to better meet industrial verification needs.

The next generation of AI systems requires strong safety guarantees. This report looks at the software implementation of neural networks and related memory safety properties, including NULL pointer deference, out-of-bound access, double-free, and memory leaks. Our goal is to detect these vulnerabilities, and automatically repair them with the help of large language models. To this end, we first expand the size of NeuroCodeBench, an existing dataset of neural network code, to about 81k programs via an automated process of program mutation. Then, we verify the memory safety of the mutated neural network implementations with ESBMC, a state-of-the-art software verifier. Whenever ESBMC spots a vulnerability, we invoke a large language model to repair the source code. For the latest task, we compare the performance of various state-of-the-art prompt engineering techniques, and an iterative approach that repeatedly calls the large language model.

ESBMC implements many state-of-the-art techniques for model checking. We report on new and improved features that allow us to obtain verification results for previously unsupported programs and properties. ESBMC employs a new static interval analysis of expressions in programs to increase verification performance. This includes interval-based reasoning over booleans and integers, forward and backward contractors, and particular optimizations related to singleton intervals because of their ubiquity. Other relevant improvements concern the verification of concurrent programs, as well as several operational models, internal ones, and also those of libraries such as pthread and the C mathematics library. An extended memory safety analysis now allows tracking of memory leaks that are considered still reachable.

Safety-critical systems with neural network components require strong guarantees. While existing neural network verification techniques have shown great progress towards this goal, they cannot prove the absence of software faults in the network implementation. This paper presents NeuroCodeBench - a verification benchmark for neural network code written in plain C. It contains 32 neural networks with 607 safety properties divided into 6 categories: maths library, activation functions, error-correcting networks, transfer function approximation, probability density estimation and reinforcement learning. Our preliminary evaluation shows that state-of-the-art software verifiers struggle to provide correct verdicts, due to their incomplete support of the standard C mathematical library and the complexity of larger neural networks., Comment: Submitted to the 2023 AFRiTS workshop

We describe and evaluate LF-checker, a metaverifier tool based on machine learning. It extracts multiple features of the program under test and predicts the optimal configuration (flags) of a bounded model checker with a decision tree. Our current work is specialised in concurrency verification and employs ESBMC as a back-end verification engine. In the paper, we demonstrate that LF-checker achieves better results than the default configuration of the underlying verification engine.

Neural networks are a powerful class of non-linear functions. However, their black-box nature makes it difficult to explain their behaviour and certify their safety. Abstraction techniques address this challenge by transforming the neural network into a simpler, over-approximated function. Unfortunately, existing abstraction techniques are slack, which limits their applicability to small local regions of the input domain. In this paper, we propose Global Interval Neural Network Abstractions with Center-Exact Reconstruction (GINNACER). Our novel abstraction technique produces sound over-approximation bounds over the whole input domain while guaranteeing exact reconstructions for any given local input. Our experiments show that GINNACER is several orders of magnitude tighter than state-of-the-art global abstraction techniques, while being competitive with local ones., Comment: Under submission to the Neural Networks Journal (revised version). Sections 2, 4.7, 5.4, Appendix A and B have been added

This work proposes a novel methodology for measuring compositional behavior in contemporary language embedding models. Specifically, we focus on adjectival modifier phenomena in adjective-noun phrases. In recent years, distributional language representation models have demonstrated great practical success. At the same time, the need for interpretability has elicited questions on their intrinsic properties and capabilities. Crucially, distributional models are often inconsistent when dealing with compositional phenomena in natural language, which has significant implications for their safety and fairness. Despite this, most current research on compositionality is directed towards improving their performance on similarity tasks only. This work takes a different approach, introducing three novel tests of compositional behavior inspired by Montague semantics. Our experimental results indicate that current neural language models do not behave according to the expected linguistic theories. This indicates that current language models may lack the capability to capture the semantic properties we evaluated on limited context, or that linguistic theories from Montagovian tradition may not match the expected capabilities of distributional models.

Neural networks are essential components of learning-based software systems. However, their high compute, memory, and power requirements make using them in low resources domains challenging. For this reason, neural networks are often quantized before deployment. Existing quantization techniques tend to degrade the network accuracy. We propose Counter-Example Guided Neural Network Quantization Refinement (CEG4N). This technique combines search-based quantization and equivalence verification: the former minimizes the computational requirements, while the latter guarantees that the network's output does not change after quantization. We evaluate CEG4N~on a diverse set of benchmarks, including large and small networks. Our technique successfully quantizes the networks in our evaluation while producing models with up to 72% better accuracy than state-of-the-art techniques.

Finding software vulnerabilities in concurrent programs is a challenging task due to the size of the state-space exploration, as the number of interleavings grows exponentially with the number of program threads and statements. We propose and evaluate EBF (Ensembles of Bounded Model Checking with Fuzzing) -- a technique that combines Bounded Model Checking (BMC) and Gray-Box Fuzzing (GBF) to find software vulnerabilities in concurrent programs. Since there are no publicly-available GBF tools for concurrent code, we first propose OpenGBF -- a new open-source concurrency-aware gray-box fuzzer that explores different thread schedules by instrumenting the code under test with random delays. Then, we build an ensemble of a BMC tool and OpenGBF in the following way. On the one hand, when the BMC tool in the ensemble returns a counterexample, we use it as a seed for OpenGBF, thus increasing the likelihood of executing paths guarded by complex mathematical expressions. On the other hand, we aggregate the outcomes of the BMC and GBF tools in the ensemble using a decision matrix, thus improving the accuracy of EBF. We evaluate EBF against state-of-the-art pure BMC tools and show that it can generate up to 14.9% more correct verification witnesses than the corresponding BMC tools alone. Furthermore, we demonstrate the efficacy of OpenGBF, by showing that it can find 24.2% of the vulnerabilities in our evaluation suite, while non-concurrency-aware GBF tools can only find 0.55%. Finally, thanks to our concurrency-aware OpenGBF, EBF detects a data race in the open-source wolfMqtt library and reproduces known bugs in several other real-world programs, which demonstrates its effectiveness in finding vulnerabilities in real-world software.

Metamorphic testing has recently been used to check the safety of neural NLP models. Its main advantage is that it does not rely on a ground truth to generate test cases. However, existing studies are mostly concerned with robustness-like metamorphic relations, limiting the scope of linguistic properties they can test. We propose three new classes of metamorphic relations, which address the properties of systematicity, compositionality and transitivity. Unlike robustness, our relations are defined over multiple source inputs, thus increasing the number of test cases that we can produce by a polynomial factor. With them, we test the internal consistency of state-of-the-art NLP models, and show that they do not always behave according to their expected linguistic properties. Lastly, we introduce a novel graphical notation that efficiently summarises the inner structure of metamorphic relations., Comment: Findings of the Association for Computational Linguistics 2022

QNNVerifier is the first open-source tool for verifying implementations of neural networks that takes into account the finite word-length (i.e. quantization) of their operands. The novel support for quantization is achieved by employing state-of-the-art software model checking (SMC) techniques. It translates the implementation of neural networks to a decidable fragment of first-order logic based on satisfiability modulo theories (SMT). The effects of fixed- and floating-point operations are represented through direct implementations given a hardware-determined precision. Furthermore, QNNVerifier allows to specify bespoke safety properties and verify the resulting model with different verification strategies (incremental and k-induction) and SMT solvers. Finally, QNNVerifier is the first tool that combines invariant inference via interval analysis and discretization of non-linear activation functions to speed up the verification of neural networks by orders of magnitude. A video presentation of QNNVerifier is available at https://youtu.be/7jMgOL41zTY, Comment: Submitted to the Demo track of the ICSE 2022 conference

Artificial Neural Networks (ANNs) are being deployed for an increasing number of safety-critical applications, including autonomous cars and medical diagnosis. However, concerns about their reliability have been raised due to their black-box nature and apparent fragility to adversarial attacks. These concerns are amplified when ANNs are deployed on restricted system, which limit the precision of mathematical operations and thus introduce additional quantization errors. Here, we develop and evaluate a novel symbolic verification framework using software model checking (SMC) and satisfiability modulo theories (SMT) to check for vulnerabilities in ANNs. More specifically, we propose several ANN-related optimizations for SMC, including invariant inference via interval analysis, slicing, expression simplifications, and discretization of non-linear activation functions. With this verification framework, we can provide formal guarantees on the safe behavior of ANNs implemented both in floating- and fixed-point arithmetic. In this regard, our verification approach was able to verify and produce adversarial examples for $52$ test cases spanning image classification and general machine learning applications. Furthermore, for small- to medium-sized ANN, our approach completes most of its verification runs in minutes. Moreover, in contrast to most state-of-the-art methods, our approach is not restricted to specific choices regarding activation functions and non-quantized representations. Our experiments show that our approach can analyze larger ANN implementations and substantially reduce the verification time compared to state-of-the-art techniques that use SMT solving., Comment: Changes with respect to the previous version: improved explanation of our methodology in Section 3; improved and extended experimental evaluation in Section 4; added comparison with the state of the art in Section 4.5

In this paper, we study influence maximization in the voter model in the presence of biased voters (or zealots) on complex networks. Under what conditions should an external controller with finite budget who aims at maximizing its influence over the system target zealots? Our analysis, based on both analytical and numerical results, shows a rich diagram of preferences and degree-dependencies of allocations to zealots and normal agents varying with the budget. We find that when we have a large budget or for low levels of zealotry, optimal strategies should give larger allocations to zealots and allocations are positively correlated with node degree. In contrast, for low budgets or highly-biased zealots, optimal strategies give higher allocations to normal agents, with some residual allocations to zealots, and allocations to both types of agents decrease with node degree. Our results emphasize that heterogeneity in agent properties strongly affects strategies for influence maximization on heterogeneous networks., Comment: 12 pages, 5 figures. This is a pre-print of a contribution published in Complex Networks XI, Springer Proceedings in Complexity 2020 (editors: Barbosa H., Gomez-Gardenes J., Gon\c{c}alves B., Mangioni G., Menezes R., Oliveira M.) published by Springer, Cham

A key challenge in crowdsourcing is inferring the ground truth from noisy and unreliable data. To do so, existing approaches rely on collecting redundant information from the crowd, and aggregating it with some probabilistic method. However, oftentimes such methods are computationally inefficient, are restricted to some specific settings, or lack theoretical guarantees. In this paper, we revisit the problem of binary classification from crowdsourced data. Specifically we propose Streaming Bayesian Inference for Crowdsourcing (SBIC), a new algorithm that does not suffer from any of these limitations. First, SBIC has low complexity and can be used in a real-time online setting. Second, SBIC has the same accuracy as the best state-of-the-art algorithms in all settings. Third, SBIC has provable asymptotic guarantees both in the online and offline settings., Comment: Accepted at the 33rd Conference on Neural Information Processing Systems (NeurIPS 2019), Vancouver, Canada

Crowdsourcing has been successfully employed in the past as an effective and cheap way to execute classification tasks and has therefore attracted the attention of the research community. However, we still lack a theoretical understanding of how to collect the labels from the crowd in an optimal way. In this paper we focus on the problem of worker allocation and compare two active learning policies proposed in the empirical literature with a uniform allocation of the available budget. To this end we make a thorough mathematical analysis of the problem and derive a new bound on the performance of the system. Furthermore we run extensive simulations in a more realistic scenario and show that our theoretical results hold in practice., Comment: paper accepted in the CrowdML workshop at NIPS 2016