Supporting Data Archive for Experimental Machine Learning Results

Zhiqi Huang provides a compressed archive of underlying data for experimental results. The archive contains two Excel files: one recording training loss data across multiple datasets and another containing detailed accuracy data from independent experimental runs. The dataset is 17.4 KB in size and was last updated on May 11, 2026.

Use Cases

• Compare training convergence across different datasets based on the loss data.
• Analyze the variance and reliability of model performance based on accuracy data from independent runs.
• Benchmark new algorithms against the provided loss and accuracy metrics.
• Reproduce or verify the experimental results described in the associated research.

Strengths

• Data is structured into dedicated files for loss and accuracy, suggesting organized experimental tracking.
• The dataset is licensed under CC-BY-4.0, permitting broad reuse with attribution.
• Contains data from multiple datasets and independent runs, allowing for comparative analysis.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is very small (17.4 KB), indicating limited scope.

Provenance

Source

Zhiqi Huang via figshare

Collection Method

Likely collected from controlled machine learning experiments.

Freshness

Last updated 2026-05-11 17:22:53; freshness should be verified.