Generative Model Reconstruction Accuracy and Output Variance by Modality and Method

A 5.5 KB Excel file uploaded to figshare on 2026-04-16 by Raffaele Marchesi. It compares the reconstruction accuracy (R²) and output variance of three generative methods—single-condition models, multi-condition models, and Coherent Denoising—across different target modalities. Each row presents the mean R² and standard deviation from 10 independent generation runs, with the best-performing method for each modality highlighted.

Use Cases

• Benchmarking generative methods based on reported reconstruction accuracy (R²) values.
• Analyzing model confidence and output variability based on the variance percentages provided.
• Comparing the performance of single-condition, multi-condition, and Coherent Denoising models for specific target modalities.

Strengths

• Provides quantitative performance metrics (mean R² ± standard deviation) derived from 10 independent runs, suggesting statistical reliability.
• Includes a secondary metric (average variance as a percentage of total real data variance) to assess model confidence and output stability.

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 (5.5 KB), indicating limited scope.

Provenance

Source

figshare

Collection Method

Likely contains results from computational experiments comparing generative models.

Time Range

Publication date is 2026-04-16; temporal coverage of the underlying experiments is unknown.

Freshness

Last updated 2026-04-16 17:49:05; freshness should be verified.

Geography

null