Poisoned MovieLens Data for Adversarial Robustness Testing

MovieLens dataset modified for adversarial machine learning and robustness testing. The specific row count, column count, and data structure are unknown.

Use Cases

• Test recommendation model robustness against adversarial attacks using poisoned user-item interaction data.
• Benchmark adversarial defense algorithms on a poisoned collaborative filtering dataset.
• Analyze the impact of poisoned data on metrics like rating prediction accuracy.

Strengths

• Designed for the specific domain of adversarial machine learning testing.
• Based on the established MovieLens recommendation dataset.

Limitations

• The size, structure, and specific poisoning methodology are unknown.
• Lacks metadata on the original data's temporal or geographic coverage.
• Cannot assess data quality, completeness, or balance without sample data.

Provenance

Collection Method

Modified from the MovieLens dataset for adversarial testing.