Saved in:
Bibliographic Details
Main Authors: Wang, Siyi, Wang, Zifan, Johanssson, Karl Henrik
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.18511
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918257490919424
author Wang, Siyi
Wang, Zifan
Johanssson, Karl Henrik
author_facet Wang, Siyi
Wang, Zifan
Johanssson, Karl Henrik
contents In interactive systems, feedback is often provided in the form of preference between queried options rather than precise scores, which motivates optimization methods to learn from such comparisons. In this work, we propose a preference-based optimization algorithm that relies on noisy two-point comparisons. At each iteration, the algorithm employs a uniform-sphere perturbation to generate a perturbed action and queries the resulting loss comparison to estimate a descent direction. We demonstrate that, under standard smoothness and bounded variance assumptions, the algorithm converges to a stationary point when the smoothing and step size parameters are properly chosen. Numerical experiments on an LQG system demonstrate the effectiveness of the preference-based optimization algorithm with comparison feedback.
format Preprint
id arxiv_https___arxiv_org_abs_2512_18511
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Preference-based optimization from noisy pairwise comparisons
Wang, Siyi
Wang, Zifan
Johanssson, Karl Henrik
Optimization and Control
In interactive systems, feedback is often provided in the form of preference between queried options rather than precise scores, which motivates optimization methods to learn from such comparisons. In this work, we propose a preference-based optimization algorithm that relies on noisy two-point comparisons. At each iteration, the algorithm employs a uniform-sphere perturbation to generate a perturbed action and queries the resulting loss comparison to estimate a descent direction. We demonstrate that, under standard smoothness and bounded variance assumptions, the algorithm converges to a stationary point when the smoothing and step size parameters are properly chosen. Numerical experiments on an LQG system demonstrate the effectiveness of the preference-based optimization algorithm with comparison feedback.
title Preference-based optimization from noisy pairwise comparisons
topic Optimization and Control
url https://arxiv.org/abs/2512.18511