Table 1
Feature extraction process.
| Step | Module | Operation | Output |
|---|---|---|---|
| 1 | SCNN | Apply convolutional, pooling, and activation layers to extract spatial structural features from the input data. | Spatial features (dimension: feature dimension/sample, unit: dimension/item) |
| 2 | BiLSTM | Perform forward and backward passes to capture temporal dependencies, then concatenate the outputs. | Temporal features (dimension: number of hidden layer units × 2 / time step, unit: dimension/step) |
| 3 | Feature Fusion | Concatenate the spatial features from SCNN and the temporal features from BiLSTM. | Fusion features (dimension: D1+D2, unit: dimension/sample) |
| 4 | Feature Optimization (IMV-GBO) | Apply the Improved Multi-variant Gradient-Based Optimization (IMV-GBO) algorithm to the fused features to dynamically optimize model parameters and minimize the loss function. | Optimized features (dimension: D1+D2, unit: dimension/sample) |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.