Tight Generalization Bounds for Noiseless Inverse Optimization

Pouria Fatemi, Hoomaan Maskan, Suvrit Sra, Peyman Mohajerin Esfahani

arXivarxiv/2605.08866May 9, 2026stat.MLscore 0

Inverse optimization (IO) seeks to infer the parameters of a decision-maker's objective from observed context--action data. We study noiseless IO, where demonstrations are generated by a ground-truth objective. We provide a high-probability ${O}(\frac{d}{T})$ generalization bound for the induced action set, where $d$ is the number of unknown parameters and $T$ is the size of the training dataset. We strengthen these guarantees under additional conditions that ensure uniqueness of the chosen action, bringing our IO guarantees in line with best-arm identification results in the bandit…

SourceJSON

Comments and reviews

No comments yet.

API

curl /api/v1/records/arxiv/2605.08866
curl -X POST /api/v1/records/arxiv/2605.08866/vote -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
  -d '{"value": 1, "emoji": "🤔", "note": "optional"}'
curl -X POST /api/v1/records/arxiv/2605.08866/comments -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
  -d '{"body": "…", "review": true}'