In response to the challenge of small and imbalanced Datasets, where the total Sample size is limited and healthy Samples significantly outweigh faulty ones, we propose a diagnostic framework designed to tackle Class imbalance, denoted as the Dual-Stream Adaptive Deep Residual Shrinkage Vision Transformer with Interclass–Intraclass Rebalancing Loss (DSADRSViT-IIRL). Firstly, to address the issue of limited Sample quantity, we incorporated the Dual-Stream Adaptive Deep Residual Shrinkage Block (DSA-DRSB) into the Vision Transformer (ViT) architecture, creating a DSA-DRSB that adaptively removes redundant signal information based on the input data characteristics. This enhancement enables the model to focus on the Global receptive field while capturing crucial local fault discrimination features from the extremely limited Samples. Furthermore, to tackle the problem of a significant Class imbalance in long-tailed Datasets, we designed an Interclass–Intraclass Rebalancing Loss (IIRL), which decouples the contributions of the Intraclass and Interclass Samples during training, thus promoting the stable convergence of the model. Finally, we conducted experiments on the Laboratory and CWRU bearing Datasets, validating the superiority of the DSADRSViT-IIRL algorithm in handling Class imbalance within mixed-load Datasets.

A labeled graph is a special structure with node identification capability, which is often used in information networks, biological networks, and other fields. The subgraph query is widely used as an important means of graph data analysis. As the size of the labeled graph increases and changes dynamically, users tend to focus on the high-match results that are of interest to them, and they want to take advantage of the relationship and number of results to get the results of the query quickly. For this reason, we consider the individual needs of users and propose a dynamic Top-K interesting subgraph query. This method establishes a novel graph topology feature index (GTSF index) including a node topology feature index (NTF index) and an edge feature index (EF index), which can effectively prune and filter the invalid nodes and edges that do not meet the restricted condition. The multi-factor candidate set filtering strategy is proposed based on the GTSF index, which can be further pruned to obtain fewer candidate sets. Then, we propose a dynamic Top-K interesting subgraph query method based on the idea of the sliding window to realize the dynamic modification of the matching results of the subgraph in the dynamic evolution of the label graph, to ensure real-time and accurate results of the query. In addition, considering the factors, such as frequent Input/Output (I/O) and network communication overheads, the optimization mechanism of the graph changes and an incremental maintenance strategy for the index are proposed to reduce the huge cost of redundant operation and global updates. The experimental results show that the proposed method can effectively deal with a dynamic Top-K interesting subgraph query on a large-scale labeled graph, at the same time the optimization mechanism of graph changes and the incremental maintenance strategy of the index can effectively reduce the maintenance overheads.