Hyperparameter Settings for Hybrid Deep Learning Framework

A hyperparameter configuration dataset supporting a novel hybrid deep learning framework for Intelligent Transportation Systems. The framework integrates BiLSTM, Transformer encoders, and Graph Convolutional Networks with API-based contextual feature fusion. The dataset, 5.5 KB in size, was authored by Mohammed Saad Javeed and last updated on 2026-05-29.

Use Cases

• Tuning hyperparameters for models combining BiLSTM, Transformer, and GCN architectures.
• Reproducing the experimental results reported for the Barcelona Accident and Metro Interstate Traffic Volume datasets.
• Benchmarking new anomaly discovery methods against the reported AUROC and AUPRC scores.
• Analyzing the impact of API fusion and GCN modeling components on model accuracy.

Strengths

• The dataset is associated with a framework achieving reported performance metrics, including an accuracy of 0.882 and an F1-score of 0.864.
• Ablation study results quantifying performance degradation (e.g., -4.1% accuracy without API fusion) are provided to validate component contributions.
• Inference latency analysis demonstrates practical deployability with a competitive accuracy-latency trade-off.

Limitations

• Row count is unknown, which may limit suitability assessment for large-scale training.
• Column-level documentation is absent; field semantics must be inferred after download.
• The description metadata is limited to the framework's performance; actual data quality requires manual inspection after download.

Provenance

Source

figshare

Collection Method

Likely generated as part of the experimental setup for a research paper proposing a hybrid deep learning framework.

Freshness

Last updated 2026-05-29 17:32:00; freshness should be verified.