HLHan Liu

Han Liu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Han Liu has a total of 24 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Time–Frequency Correlation-Aware Few-Shot Learning for Time-Series Anomaly Detection

    by Liang Xi, Runze Li, Han Liu, Chen Chen, Yuan Cheng

    Time Series Anomaly Detection (TSAD) often leverages unsupervised learning to handle extensive sensor-generated unlabeled data. However, existing methods ignore to utilize the limited labeled anomalies, and overlook the correlations between data in both the time and frequency domains, resulting in the inability to adequately capture the discrepancies between normal and anomalous patterns. To fully leverage on the exemplary role of labeled anomalies and maximize the representation learning ability, we propose a weakly-supervised Time-Frequency Correlation aware Few-Shot Learning method for TSAD (TFC-FSL). Specifically, we design a few-shot learning strategy and construct correlation graphs in time-domain and frequency-domain, respectively, to fully mine the latent information in time series. Then, we employ an autoencoder framework to extract the original and correlation representations in both time and frequency domains, and fuse them with the reconstruction representations. Finally, we design an estimation network with the guiding of few-shot anomalies to amplify the normal/abnormal distribution discrepancies for TSAD tasks. Experimental results show that TFC-FSL outperforms the other models.

  • DuDi: Dual-Scope Discrepancy-Aware Attributed Network Anomaly Detection

    by Liang Xi, Runze Li, Dichen Li, Ruidong Wang, Han Liu, Huiying Xu, Xinzhong Zhu

    In attributed networks, capturing the correlation between different scopes is crucial to detect anomalies, but few methods focus on modeling the correlation discrepancy of normal and abnormal nodes between different scopes for anomaly detection. Further analysis shows that normal nodes show strong correlation between global and local scopes, while abnormal nodes show weak correlation between the two scopes. Inspired by this perspective, we present a Dual-scope Discrepancy-aware attributed network anomaly detection method (DuDi), to capture macro normal patterns of nodes while reducing noise perturbations based on graph diffusion convolution, and micro patterns of nodes with local graph sampling strategies, respectively, resulting in weak correlation between macro and micro scopes for abnormal nodes, while strong correlation between the two scopes for normal nodes. Based on the dual-scope correlation discrepancy of normal and abnormal nodes, we design a discrepancy-aware anomaly score with Canonical Correlation Analysis (CCA) for anomaly detection. Comprehensive experiments conducted on four real-world datasets showcase the efficacy of DuDi compared to state-of-the-art methods. The experiments show that DuDi achieves significant performance improvements, with up to 6.14% improvement in AUC and 10.33% improvement in AP over the best baselines across multiple datasets.

  • Multivariate Time Series Anomaly Detection With Hierarchical Component-Aware

    by Han Liu, Zheng Zhang, Liang Xi, Fengbin Zhang

    Multivariate time-series anomaly detection (MTSAD) remains challenging due to the complexity of spatiotemporal dependencies, nonstationary dynamics, and heterogeneous variable interactions. Existing methods often struggle to simultaneously model correlations within and across different time series components, limiting their ability to capture hierarchical patterns at multiple scales. To address these issues, we propose hierarchical component-aware multivariate time-series anomaly detection (HCAAD), an unsupervised framework that combines frequency-adaptive multiscale decomposition with cross-component correlation modeling. First, we use a fast Fourier transform (FFT)-based decomposition to split the time series into multiple components. This step isolates long-term trends, seasonal cycles, and transient fluctuations. Second, we design a dynamic correlation matrix to explicitly model intra- and intercomponent dependencies. An attention mechanism further refines these correlations by adaptively integrating global spatiotemporal patterns. Experiments on six benchmark datasets show that HCAAD consistently achieves state-of-the-art performance.


  • MGTC: Multi-Granularity Temporal Aware Time Series Classification

    by Liang Xi, Wei Wang, Han Liu, Jingqi Pan

    Self-supervised learning (SSL) can extract useful temporal representations for time series classification (TSC) tasks. However, existing methods with subsequence and instance-level augmentation lead to the loss of global information and inductive bias. Meanwhile, neglecting multi-granularity temporal representations also poses challenges for modeling complex temporal structures and relationships. Therefore, in this work, we propose a multi-granularity temporal-aware time series classification method, called MGTC, that learns multi-granularity temporal representations, enhancing the ability to perceive class discrepancies. Specifically, we propose a multi-density masking strategy to adapt the dynamic time-varying patterns for learning comprehensive temporal representations. Next, we employ cluster-wise constraint to hierarchically aggregate these representations at the instance level. Finally, we design two self-supervised tasks: i) granularity-aware contrastive learning, to extract intra-instance fine-grained temporal structures and inter-instance coarse-grained class relationships, and ii) cross-view prediction pretext task, to capture global contextual temporal dependencies. We conducted comprehensive experimental evaluations on various types of dataset, and the experimental results validated the effectiveness of our method. The code is publicly available at https://github.com/mrxiliang/MGTC.


  • OpenFi: Open-Set WiFi Human Sensing via Virtual Embedding Confidence-Aware

    by Han Liu, Liang Xi, Wei Wang, Fengbin Zhang, Zygmunt J. Haas

    WiFi sensing technology utilizes Channel State Information (CSI) to analyze human behavior and plays a crucial role in mobile computing. WiFi sensing systems are typically deployed by collecting data in specific known environments, known as closed-set settings. However, in practical deployment, WiFi sensing systems may encounter unknown environments, generating unknown CSI patterns due to signal reflection, multipath effects, and interference. In such open-set conditions, the WiFi sensing system should possess the capability to recognize unknown CSI patterns, enhancing its security and reliability. In response, this work proposes an open-set WiFi human sensing method based on virtual embedding confidence-aware (OpenFi). The core of OpenFi is virtual embedding generation to simulate a realistic open-set feature space. This strategy minimizes both empirical and open-set risks, enabling OpenFi to recognize unknown CSI patterns effectively. We conducted extensive experiments on diverse datasets, covering various WiFi sensing tasks, including human identification, human activity recognition, and sign language recognition. Experimental results demonstrate that OpenFi accurately identifies previously unseen CSI patterns in open-set conditions, achieving significant improvements of up to 27% and 10.26% in the FPR95 and OSCR metrics, respectively


  • ExTAD: Embedding Exchange Inspired Time Series Anomaly Detection With Modal Consistency

    by Han Liu, Liang Xi, Minghao Gu, Sizhe Huang, Chaoyang Sheng, Fengbin Zhang

    Comprehensive learning of temporal representations is crucial for time series anomaly detection (TSAD). Frequency domain analysis has been proven to be an effective strategy for detecting diverse patterns of anomalies. However, existing methods ignore the exploration of complementary information of modalities and the anomaly misjudgments caused by time-frequency distribution shifts. In this paper, we propose Embedding exchange-inspired unsupervised Time-series Anomaly Detection (ExTAD) with modal consistency to learn multimodal representations that improve detection performance. Specifically, we employ a weight-sharing dual-AutoEncoder(dual-AE) to extract the time-frequency embedding. Then, embedding exchange is used to learn robust perturbation-invariant representations for miningmultimodal complementary information. Furthermore, we design a modal consistency strategy to align time-frequency representations to a unified scale, alleviating anomaly misjudgments caused by distribution shifts. Finally, we combine modal consistency loss and reconstruction errors to perform time series anomaly detection tasks. Experimental results on six benchmark datasets show that ExTAD achieves state-of-the-art performance compared to existing methods.


  • Bidirectional consistency with temporal-aware for semi-supervised time series classification

    by Han Liu, Fengbin Zhang, Xunhua Huang, Ruidong Wang, Liang Xi


  • KalmanAE: Deep Embedding Optimized Kalman Filter for Time Series Anomaly Detection

    by Xunhua Huang, Fengbin Zhang, Ruidong Wang, Xiaohui Lin, Han Liu, Haoyi Fan

    The Kalman filter performs well in system state estimation by inferring a joint probability distribution over time variables, which has numerous technological applications in time series analysis. However, the complex Kalman filter parameter settings prevent it from optimally estimating the system state, and this suboptimal estimation makes it difficult to effectively distinguish between normal and anomalous system states in anomaly detection. In this paper, we propose a deep embedding-optimized Kalman filter for unsupervised time series anomaly detection, where the system state of a normal time series can be fitted by the embedding-optimized Kalman filter in an unsupervised manner and anomalies can be detected from data points that deviate from the normal system state. Specifically, we use an autoencoder-enhanced Kalman filter to capture the normal pattern of the time series, where the original time series signal is first fed into the Kalman filter, then the autoencoder encodes the filtered signal and reconstructs the original signal, and the learned embedding from encoder is used to sequentially optimize the Kalman filter. Finally, the optimized filter captures the normal pattern of the time series and the reconstruction error from the filtered signal can be measured to detect anomalies. The validity of the method is verified on real-world time series datasets.