Multivariate time series anomaly detection is a crucial factor in real-world applications but a challenging task due to the complex temporal dependencies and system dynamics. Reconstruction-based methods have made great improvements in recent years. However, we observe an issue these methods are suffering, that they primarily measure deviations in the time points themselves when performing anomaly detection but ignore changes in the dynamic properties of the system. In these cases, they are unable to produce sufficient reconstruction errors to detect anomalies, so some potential abnormal time points caused by the dynamic evolution of the system are missing. To address this problem, we propose a novel method, SDA2D, which models system dynamics by the derivative of the NCDE-derived state vector with respect to time, enabling the learning of reconstruction deviation and system evolution jointly. Our experimental results show that SDA2D achieves noticeable improvements in four benchmark datasets, and the visualization also provides further instructions for anomaly diagnosis, which helps locate the sources of these anomalies.

Reconstruction-based methods are competitive choices for multivariate time series anomaly detection (MTS AD). However, one challenge these methods may suffer is over generalization, where abnormal inputs are also well reconstructed. In addition, balancing robustness and sensitivity is also important for final performance, as robustness ensures accurate detection in potentially noisy data, while sensitivity enables early detection of subtle anomalies. To address these problems, inspired by idempotent generative network, we take the view from the manifold and propose a novel module named **I**dempotent **G**eneration for **A**nomaly **D**etection (IGAD) which can be flexibly combined with a reconstruction-based method without introducing additional trainable parameters. We modify the manifold to make sure that normal time points can be mapped onto it while tightening it to drop out abnormal time points simultaneously. Regarding the latest findings of AD metrics, we evaluated IGAD on various methods with four real-world datasets, and they achieve visible improvements in VUS-PR than their predecessors, demonstrating the effective potential of IGAD for further improvements in MTS AD tasks. Our instructions on integrating IGAD into customized models and example codes are available at https://github.com/ProEcho1/Idempotent-Generation-for-Anomaly-Detection-IGAD.

Proteins serve as the functional building blocks of life, facilitating critical tasks such as signaling, catalysis, and structural support in all living organisms. Designing proteins with targeted biological features or domains is of utmost importance. Traditional wet-lab experiments are time-consuming and resource-intensive, which makes deep learning (DL) methods ideal alternatives. However, existing DL methods predominantly focus on generating new proteins with the same biological domain as the training data, and overlook some scenarios where designers expect to combine proteins from different biological domains to create novel proteins with both features, which can show better fits for practical purpose. To fill this gap, in this paper, we present ComProtein, a novel framework further exploiting the potential of pre-trained protein large language models, which is the first work aiming to generate innovative proteins with combinative biological features from two different domains. This process is performed by a cycle-consistent generative adversarial approach, leveraging insights from the latent space. It enables the transformation of protein representations from one biological domain to another, while preserving their intrinsic features. Additionally, we introduce new evaluative metrics, namely Shortest Target Neighbor Distance (STND), Mutual Root Mean Square Deviation (MRMSD) and Sequence Diversity (SD) on the evaluation of biological representations, protein structure and sequence quality, respectively to complement the existing measures. Our experimental results demonstrate that our proposed method performs better and has great potential in biological representations, structure similarity, homology relationships, and sequence quality.