Time series data naturally exist in many domains including medical data analysis, infrastructure sensor monitoring, and motion tracking. However, a very small portion of anomalous time series can be observed, comparing to the whole data. Most existing approaches are based on the supervised classification model requiring representative labels for anomaly class(es), which is challenging in real-world problems. So can we learn how to detect anomalous time ticks in an effective yet efficient way, given mostly normal time series data? Therefore, we propose an unsupervised reconstruction model named BeatGAN which learns to detect anomalies based on normal data, or data which majority of samples are normal. BeatGAN provides a framework to adversarially learn to reconstruct, which can cooperate with both 1-d CNN and RNN. Rarely observed anomalies can result in larger reconstruction errors, which are then detected based on extreme value theory. Moreover, data augmentation with dynamic time warping regularizes reconstruction and provides robustness. In the experiments, effectiveness and sensitivity are studied in both synthetic data and various real-world time series. BeatGAN achieves better accuracy and fast inference.
Cumulative effect in social contagion underlies many studies on the spread of innovation, behavior, and influence. However, few large-scale empirical studies are conducted to validate the existence of cumulative effect in information diffusion on social networks. In this paper, using the population-scale dataset from the largest Chinese microblogging website, we conduct a comprehensive study on the cumulative effect in information diffusion. We base our study on the diffusion network of message, where nodes are the involved users and links characterize forwarding relationship among them. We find that multiple exposures to the same message indeed increase the possibility of forwarding it. However, additional exposures cannot further improve the chance of forwarding when the number of exposures crosses its peak at two. This finding questions the cumulative effect hypothesis in information diffusion. Furthermore, to clarify the forwarding preference among users, we investigate both structural motif in the diffusion network and temporal pattern in information diffusion process. Findings provide some insights for understanding the variation of message popularity and explain the characteristics of diffusion network.
In this paper, we consider the problem of exploring structural regularities of networks by dividing the nodes of a network into groups such that the members of each group have similar patterns of connections to other groups. Specifically, we propose a general statistical model to describe network structure. In this model, a group is viewed as a hidden or unobserved quantity and it is learned by fitting the observed network data using the expectation-maximization algorithm. Compared with existing models, the most prominent strength of our model is the high flexibility. This strength enables it to possess the advantages of existing models and to overcome their shortcomings in a unified way. As a result, not only can broad types of structure be detected without prior knowledge of the type of intrinsic regularities existing in the target network, but also the type of identified structure can be directly learned from the network. Moreover, by differentiating outgoing edges from incoming edges, our model can detect several types of structural regularities beyond competing models. Tests on a number of real world and artificial networks demonstrate that our model outperforms the state-of-the-art model in shedding light on the structural regularities of networks, including the overlapping community structure, multipartite structure, and several other types of structure, which are beyond the capability of existing models.
Making the contents generated by Large Language Model (LLM), accurate, credible and traceable is crucial, especially in complex knowledge-intensive tasks that require multi-step reasoning and each step needs knowledge to solve. Retrieval-augmented generation is good potential to solve this problem. However, where and how to introduce Information Retrieval (IR) to LLM is a big challenge. Previous work has the problems that wrong knowledge retrieved by IR misleads the LLM and interaction between IR and LLM breaks the reasoning chain of LLM. This paper proposes a novel framework named Search-in-the-Chain (SearChain) for the interaction between LLM and IR to solve the challenges. First, LLM generates the reasoning chain named Chain-of-Query (CoQ) where each node consists of an IR-oriented query-answer pair. Second, IR verifies the answer of each node of CoQ. It corrects the answer that is not consistent with the retrieved information when IR gives high confidence, which improves the credibility. Third, LLM can indicate its missing knowledge in CoQ and rely on IR to provide this knowledge to LLM. These operations improve the accuracy in terms of reasoning and knowledge. Finally, SearChain generates the reasoning process and marks references to supporting documents for each reasoning step, which improves traceability. Interaction with IR in SearChain forms a novel reasoning path based on a tree, which enables LLM to dynamically modify the direction of reasoning. Experiments show that SearChain outperforms state-of-the-art baselines on complex knowledge-intensive tasks including multi-hop Q&A, slot filling, fact checking, and long-form Q&A.
Shicheng Xu, Liang Pang, Mo Yu, Fandong Meng, Huawei Shen, Xueqi Cheng, Jie Zhou. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.
1 June 2012 in Physica A: Statistical Mechanics and its Applications