Craven and Shavlik: extracting tree-structured representations of trained networks

What changed

Presented TREPAN, which queries a trained neural network to induce a compact decision tree intended to approximate the network while making its learned behavior more comprehensible.

Scientific scope

The record is the exact NIPS 1995 paper. It does not establish universal faithfulness or safety, and later explanation methods and interpretability benchmarks remain separate.

Evidence

The linked sources identify this contribution and its scope. WRS has not independently reconstructed the scientific claims. The record is the exact NIPS 1995 paper. It does not establish universal faithfulness or safety, and later explanation methods and interpretability benchmarks remain separate.

Contributors

• Mark W. Craven · SCIENCE: Named source author of the scoped work; individual effort shares are not measured.
• Jude W. Shavlik · SCIENCE: Named source author of the scoped work; individual effort shares are not measured.

• NeurIPS record for TREPAN paper
• TREPAN paper full text

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction