Kunxun Qi, Jianfeng Du, Hai Wan. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Anomaly detection aims to detect instances that deviate significantly from the majority. Due to the difficulties of collecting a large amount of anomalies in practice, existing methods generally assume the availability of a clean normal dataset and leverage it to detect anomalies by characterizing the normality of normal samples. However, for many application scenarios, collecting a normal dataset that is sufficiently clean is not easy. What is often observed is that a small amount of anomalies are often falsely mixed into the normal dataset, resulting in a contaminated dataset. Obviously, the contamination in the normal dataset could significantly compromise the model's ability to detect anomalies. To alleviate this issue, two contamination-immune bidirectional generative adversarial networks (BiGAN) are developed, which can learn the probability distribution of normal samples from a contaminated dataset under some mild conditions. Rigorous proofs are provided to guarantee the theoretical correctness of the proposed models. Thanks to the removing of negative influences from the contamination samples, the proposed contamination-immune models can thus be applied to detect anomalies accurately for the scenarios with contaminated datasets. Extensive experimental results show that the proposed method outperforms the current state-of-the-art (SOTA) ones significantly under the scenarios with contaminated training datasets.

Aspect-based sentiment analysis (ABSA) has attracted much attention due to its wide application scenarios. Most previous studies have focused solely on monolingual ABSA, posing a formidable challenge when extending ABSA applications to multilingual scenarios. In this paper, we study upgrading monolingual ABSA to cross-lingual ABSA. Existing methods usually exploit pre-trained cross-lingual language to model cross-lingual ABSA, and enhance the model with translation data. However, the low-resource languages might be under-represented during the pre-training phase, and the translation-enhanced methods heavily rely on the quality of the translation and label projection. Inspired by the observation that quantum entanglement can correlate multiple single systems, we map the monolingual expression to the quantum Hilbert space as a single quantum system, and then utilize quantum entanglement and quantum measurement to achieve cross-lingual ABSA. Specifically, we propose a novel quantum neural model named QPEN (short for quantum projection and quantum entanglement enhanced network). It is equipped with a proposed quantum projection module that projects aspects as quantum superposition on a complex-valued Hilbert space. Furthermore, a quantum entanglement module is proposed in QPEN to share language-specific features between different languages without transmission. We conducted simulation experiments on the classical computer, and experimental results on SemEval-2016 dataset demonstrate that our method achieves state-of-the-art performance in terms of F1-scores for five languages.

It is important to automatically discover the underlying tree-structured formulae from large amounts of data. In this paper, we examine learning linear temporal logic on finite traces (LTLf) formulae, which is a tree structure syntactically and characterizes temporal properties semantically. Its core challenge is to bridge the gap between the concise tree-structured syntax and the complex LTLf semantics. Besides, the learning quality is endangered by explosion of the search space and wrong search bias guided by imperfect data. We tackle these challenges by proposing an LTLf encoding method to parameterize a neural network so that the neural computation is able to simulate the inference of LTLf formulae. We first identify faithful LTLf encoding, a subclass of LTLf encoding, which has a one-to-one correspondence to LTLf formulae. Faithful encoding guarantees that the learned parameter assignment of the neural network can directly be interpreted to an LTLf formula. With such an encoding method, we then propose an end-to-end approach, TLTLf, to learn LTLf formulae through neural networks parameterized by our LTLf encoding method. Experimental results demonstrate that our approach achieves state-of-the-art performance with up to 7% improvement in accuracy, highlighting the benefits of introducing the faithful LTLf encoding.

Linear temporal logic (LTL) satisfiability checking is a fundamental and hard (PSPACE-complete) problem. In this paper, we explore checking LTL satisfiability via end-to-end learning, so that we can take only polynomial time to check LTL satisfiability. Existing approaches have shown that it is possible to leverage end-to-end neural networks to predict the Boolean satisfiability problem with performance considerably higher than random guessing. Inspired by these approaches, we study two interesting questions: can end-to-end neural networks check LTL satisfiability, and can neural networks capture the semantics of LTL? To this end, we train different neural networks for keeping three logical properties of LTL, i.e., recursive property, permutation invariance, and sequentiality. We demonstrate that neural networks can indeed capture some effective biases for checking LTL satisfiability. Besides, designing a special neural network keeping the logical properties of LTL can provide a better inductive bias. We also show the competitive results of neural networks compared with state-of-the-art approaches, i.e., nuXmv and Aalta, on large scale datasets.

Fault tree analysis (FTA) is a prominent reliability analysis method, which is widely used in safety-critical industries. Computing the minimal cut sets (MCSs) of a fault tree, i.e., finding all the smallest combinations of the basic events that cause system failures, is a fundamental step in FTA. Since coherent fault trees are the most common in industrial systems in practice, they are the focus of this article. Computing MCSs is a computationally hard problem. Classical methods have been proposed based on manipulation of Boolean expressions and binary decision diagrams. However, given the inherent intractability of computing MCSs in practice, there are still limitations on time and memory in these methods. Therefore, developing new methods over different paradigms remains to be an interesting research direction. In this article, motivated by recent progress on modern Boolean satisfiability problem (SAT) solvers, we present a new method for computing MCSs based on SAT, namely SATMCS. Specifically, given a fault tree, we iteratively search for a cut set based on the conflict-driven clause learning framework. By exploiting local propagation graph, which characterizes the partial failure propagation based on the cut set, we provide efficient algorithms for extracting an MCS. The new MCS is learned as a block clause for SAT solving, and the conflict clauses in iterations are incrementally recorded, which helps to prune search space and ensures completeness of the results. Moreover, we adopt a jump-chronological backtracking strategy to prepare the next iteration, which allows for reusing the same search steps in SAT solving. We compare SATMCS with state-of-the-art commercial tools on practical fault trees. Although SATMCS is only a prototype, it shows comparable performance in time consumption with one tool (XFTA), and in various cases, it outperforms the others (FaultTree+ and Commander). Besides, SATMCS exhibits much better performance on memory usage than these tools. Specifically, SATMCS consumes about one order of magnitude less memory usage in most instances.