Real-World Robustness Evaluation Results for Emotion Recognition Models

Weiguang Dong's experimental results for a hybrid emotion recognition model, published on figshare in April 2026. The 5.5 KB XLS file contains performance metrics from evaluations on three benchmark datasets: SemEval-2018, RAVDESS, and CMU-MOSEI.

Use Cases

• Benchmarking new emotion recognition models against reported state-of-the-art accuracy and F1-scores.
• Analyzing model robustness across text-only and multimodal data scenarios based on the described cross-modal generalization.
• Informing the design of affective computing systems based on performance results from established datasets.

Strengths

• Results are benchmarked against three established datasets: SemEval-2018, RAVDESS, and CMU-MOSEI.
• Specific performance metrics are provided, including 58.5% accuracy and 68.7% F1-score on SemEval-2018.
• The dataset is small (5.5 KB) and likely focused, facilitating quick inspection.

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 tiny, containing only evaluation results, not the underlying training or test data.

Provenance

Source

Weiguang Dong

Collection Method

Likely contains aggregated experimental results from model evaluation.

Time Range

null

Freshness

Last updated 2026-04-27 17:22:33; freshness should be verified.

Geography

null