Safe reinforcement learning (SRL) aims to realize a safe learning process for deep reinforcement learning (DRL) algorithms by incorporating safety constraints. However, the efficacy of SRL approaches often relies on accurate function approximations, which are notably challenging to achieve in the early learning stages due to data insufficiency. To address this issue, we introduce, in this work, a novel generalizable safety enhancer (GenSafe) that can overcome the challenge of data insufficiency and enhance the performance of SRL approaches. Leveraging model order reduction techniques, we first propose an innovative method to construct a reduced order Markov decision process (ROMDP) as a low-dimensional approximator of the original safety constraints. Then, by solving the reformulated ROMDP-based constraints, GenSafe refines the actions of the agent to increase the possibility of constraint satisfaction. Essentially, GenSafe acts as an additional safety layer for SRL algorithms. We evaluate GenSafe on multiple SRL approaches and benchmark problems. The results demonstrate its capability to improve safety performance, especially in the early learning phases, while maintaining satisfactory task performance. Our proposed GenSafe not only offers a novel measure to augment existing SRL methods but also shows broad compatibility with various SRL algorithms, making it applicable to a wide range of systems and SRL problems.

Over the past decade, Artificial Intelligence (AI) has had great success recently and is being used in a wide range of academic and industrial fields. More recently, Large Language Models (LLMs) have made rapid advancements that have propelled AI to a new level, enabling and empowering even more diverse applications and industrial domains with intelligence, particularly in areas like software engineering and natural language processing. Nevertheless, a number of emerging trustworthiness concerns and issues exhibited in LLMs, e.g., robustness and hallucination, have already recently received much attention, without properly solving which the widespread adoption of LLMs could be greatly hindered in practice. The distinctive characteristics of LLMs, such as the self-attention mechanism, extremely large neural network scale, and autoregressive generation usage contexts, differ from classic AI software based on Convolutional Neural Networks and Recurrent Neural Networks and present new challenges for quality analysis. Up to the present, it still lacks universal and systematic analysis techniques for LLMs despite the urgent industrial demand across diverse domains. Towards bridging such a gap, in this paper, we initiate an early exploratory study and propose a universal analysis framework for LLMs, namedLUNA, which is designed to be general and extensible and enables versatile analysis of LLMs from multiple quality perspectives in a human-interpretable manner. In particular, we first leverage the data from desired trustworthiness perspectives to construct an abstract model as an auxiliary analysis asset and proxy, which is empowered by various abstract model construction methods built-inLUNA. To assess the quality of the abstract model, we collect and define a number of evaluation metrics, aiming at both the abstract model level and the semantics level. Then, the semantics, which is the degree of satisfaction of the LLM w.r.t. the trustworthiness perspective, is bound to and enriches the abstract model with semantics, which enables more detailed analysis applications for diverse purposes, e.g., abnormal behavior detection. To better understand the potential usefulness of our analysis frameworkLUNA, we conduct a large-scale evaluation, the results of which demonstrate that 1) the abstract model has the potential to distinguish normal and abnormal behavior in LLM, 2)LUNAis effective for the real-world analysis of LLMs in practice, and the hyperparameter settings influence the performance, 3) different evaluation metrics are in different correlations with the analysis performance. In order to encourage further studies in the quality assurance of LLMs, we made all of the code and more detailed experimental results data available on the supplementary website of this paperhttps://sites.google.com/view/llm-luna.

Deep neural networks (DNNs) have been widely adopted in many decision-making industrial applications. Their fairness issues, i.e., whether there exist unintended biases in the DNN, receive much attention and become critical concerns, which can directly cause negative impacts in our daily life and potentially undermine the fairness of our society, especially with their increasing deployment at an unprecedented speed. Recently, some early attempts have been made to provide fairness assurance of DNNs, such as fairness testing, which aims at finding discriminatory samples empirically, and fairness certification, which develops sound but not complete analysis to certify the fairness of DNNs. Nevertheless, how to formally compute discriminatory samples and fairness scores (i.e., the percentage of fair input space), is still largely uninvestigated. In this paper, we propose DeepGemini, a novel fairness formal analysis technique for DNNs, which contains two key components: discriminatory sample discovery and fairness score computation. To uncover discriminatory samples, we encode the fairness of DNNs as safety properties and search for discriminatory samples by means of state-of-the-art verification techniques for DNNs. This reduction enables us to be the first to formally compute discriminatory samples. To compute the fairness score, we develop counterexample guided fairness analysis, which utilizes four heuristics to efficiently approximate a lower bound of fairness score. Extensive experimental evaluations demonstrate the effectiveness and efficiency of DeepGemini on commonly-used benchmarks, and DeepGemini outperforms state-of-the-art DNN fairness certification approaches in terms of both efficiency and scalability.

Abstract This paper presents a property-directed approach to verifying recurrent neural networks (RNNs). To this end, we learn a deterministic finite automaton as a surrogate model from a given RNN using active automata learning. This model may then be analyzed using model checking as a verification technique. The term property-directed reflects the idea that our procedure is guided and controlled by the given property rather than performing the two steps separately. We show that this not only allows us to discover small counterexamples fast, but also to generalize them by pumping toward faulty flows hinting at the underlying error in the RNN. We also show that our method can be efficiently used for adversarial robustness certification of RNNs.

Cyber-Physical Systems (CPSs) have been widely adopted in various industry domains to support many important tasks that impact our daily lives, such as automotive vehicles, robotics manufacturing, and energy systems. As Artificial Intelligence (AI) has demonstrated its promising abilities in diverse tasks like decision-making, prediction, and optimization, a growing number of CPSs adopt AI components in the loop to further extend their efficiency and performance. However, these modern AI-enabled CPSs have to tackle pivotal problems that the AI-enabled control systems might need to compensate the balance acrossmultiple operation requirementsand avoid possible defections in advance to safeguard human lives and properties. Modular redundancy and ensemble method are two widely adopted solutions in the traditional CPSs and AI communities to enhance the functionality and flexibility of a system. Nevertheless, there is a lack of deep understanding of the effectiveness of such ensemble design on AI-CPSs across diverse industrial applications. Considering the complexity of AI-CPSs, existing ensemble methods fall short of handling such huge state space and sophisticated system dynamics. Furthermore, an ideal control solution should consider the multiple system specifications in real-time and avoid erroneous behaviors beforehand. Such that, a new specification-oriented ensemble control system is of urgent need for AI-CPSs. In this paper, we propose$\mathtt {SIEGE}$, a semantics-guided ensemble control framework to initiate an early exploratory study of ensemble methods on AI-CPSs and aim to construct an efficient, robust, and reliable control solution for multi-tasks AI-CPSs. We first utilize a semantic-based abstraction to decompose the large state space, capture the ongoing system status and predict future conditions in terms of the satisfaction of specifications. We propose a series of new semantics-aware ensemble strategies and an end-to-end Deep Reinforcement Learning (DRL) hierarchical ensemble method to improve the flexibility and reliability of the control systems. Our large-scale, comprehensive evaluations over five subject CPSs show that 1) the semantics abstraction can efficiently narrow the large state space and predict the semantics of incoming states, 2) our semantics-guided methods outperform state-of-the-art individual controllers and traditional ensemble methods, and 3) the DRL hierarchical ensemble approach shows promising capabilities to deliver a more robust, efficient, and safety-assured control system. To enable further research along this direction to build better AI-enabled CPS, we made all of the code and experimental results data publicly available athttps://sites.google.com/view/ai-cps-siege/home.

Formal verification has emerged as a powerful approach to ensure the safety and reliability of deep neural networks. However, current verification tools are limited to only a handful of properties that can be expressed as first-order constraints over the inputs and output of a network. While adversarial robustness and fairness fall under this category, many real-world properties (e.g., "an autonomous vehicle has to stop in front of a stop sign") remain outside the scope of existing verification technology. To mitigate this severe practical restriction, we introduce a novel framework for verifying neural networks, named neuro-symbolic verification. The key idea is to use neural networks as part of the otherwise logical specification, enabling the verification of a wide variety of complex, real-world properties, including the one above. A defining feature of our framework is that it can be implemented on top of existing verification infrastructure for neural networks, making it easily accessible to researchers and practitioners.

Online monitoring is a popular validation approach in which the temporal behavior of a system is checked to assess whether it satisfies a given specification expressed, e.g., in signal temporal logic (STL). This is done by employing a monitor that, at each time point, states the specification validity: satisfied, violated, or unknown. In some settings, monitoring should continue even after a violation episode is detected, to detect possible future violation episodes. However, for a monitor just relying on STL semantics, this is not possible, as, once the specification is violated by an input signal, any continuation of the signal still violates the specification. To tackle this problem, we here propose an optimal reset technique that, at runtime, detects the end of a violation episode and shifts the evaluation of the monitor to skip such an episode. In this way, the monitoring can continue to detect possible other future violation episodes. We propose a framework that integrates the reset technique with an existing monitoring approach. Experiments on two Simulink models show that the technique can effectively reset the monitor and report all the violation episodes, with a negligible overhead on the monitoring cost.

Misuses of library APIs are pervasive and often lead to software crashes and vulnerability issues. Various static analysis tools have been proposed to detect library API misuses. They often involve mining frequent patterns from a large number of correct API usage examples, which can be hard to obtain in practice. They also suffer from low precision due to an over-simplified assumption that a deviation from frequent usage patterns indicates a misuse. We make two observations on the discovery of API misuse patterns. First, API misuses can be represented as mutants of the corresponding correct usages. Second, whether a mutant will introduce a misuse can be validated via executing it against a test suite and analyzing the execution information. Based on these observations, we propose MutApi, the first approach to discovering API misuse patterns via mutation analysis. To effectively mimic API misuses based on correct usages, we first design eight effective mutation operators inspired by the common characteristics of API misuses. MutApi generates mutants by applying these mutation operators on a set of client projects and collects mutant-killing tests as well as the associated stack traces. Misuse patterns are discovered from the killed mutants that are prioritized according to their likelihood of causing API misuses based on the collected information. We applied MutApi on 16 client projects with respect to 73 popular Java APIs. The results show that MutApi is able to discover substantial API misuse patterns with a high precision of 0.78. It also achieves a recall of $0.49$ on the MuBench benchmark, which outperforms the state-of-the-art techniques.

Bug-inducing commits provide important information to understand when and how bugs were introduced. Therefore, they have been extensively investigated by existing studies and frequently leveraged to facilitate bug fixings in industrial practices.