Table 1_Formal methods for safety-critical machine learning: a systematic literature revie

46 peer-reviewed studies from 2020 to mid-2025 are synthesized in this systematic literature review by Alexandra Newcomb. The work identifies and classifies research into eight methodological categories for applying formal methods to machine learning safety. It serves as a state-of-the-art reference and roadmap for advancing the safe deployment of ML systems.

Use Cases

• Surveying methodological advances in formal verification for ML based on the eight identified categories.
• Identifying research gaps and future opportunities in ML safety verification based on the compiled analysis.
• Benchmarking new verification techniques against the synthesized state-of-the-art from 46 studies.
• Informing the development of safety standards for ML in critical domains based on the review of application areas.

Strengths

• Structured review protocol identified 46 studies from four major digital libraries.
• Classifies research into eight distinct methodological categories for clarity.
• Covers peer-reviewed literature from 2020 to mid-2025, providing a recent snapshot.

Limitations

• The 674.8 KB file size suggests the dataset is a document, not a structured data table; analysis requires manual extraction.
• Row count and column-level documentation are absent; data semantics must be inferred from the review text.
• The dataset's scope is limited to the literature reviewed and may not be exhaustive of all relevant work.

Provenance

Source

Alexandra Newcomb via figshare

Collection Method

Systematic Literature Review following a structured protocol.

Time Range

2020 to mid-2025

Freshness

Last updated 2026-03-17 13:07:08