A table summarizing total parameter counts, epoch training times, and epoch inference times for various deep learning models and Inception-ResNetV2 variants integrated with attention mechanisms. Author Chao Zhang uploaded the 9.4 KB XLSX file to figshare, where it was last updated on May 14, 2026. The data highlights a trade-off between model complexity and computational efficiency, noting that an SE-enhanced variant had 55.0M parameters and a 17-second training time per epoch.
Use Cases
- Compare computational cost across different deep learning architectures based on reported parameter counts and training times.
- Evaluate the inference time impact of integrating attention mechanisms like SE modules into a base model.
- Assess the feasibility of model deployment in clinical settings based on the reported ~1 second per epoch inference time.
Strengths
- Includes specific, directly comparable metrics such as a 55.0M parameter count for one model variant.
- Provides concrete timing data, including a 17-second training time per epoch and a ~1-second inference time per epoch.
- Explicitly documents the trade-off between model complexity and computational efficiency for clinical deployment.
Limitations
- Row count is unknown, which may limit suitability assessment.
- Column-level documentation is absent; field semantics must be inferred after download.
- The dataset is very small at 9.4 KB, indicating limited scope.
Provenance
- Source
- Chao Zhang via figshare.
- Freshness
- Last updated 2026-05-14 17:44:29; freshness should be verified.