Research map / Large language models / Alignment and preferences
Supporting High-Stakes Decision Making Through Interactive Preference Elicitation in the Latent Space
Michael Eichelbeck, Tim Voigt, Matthias Althoff
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| run PBO in the autoencoder latent space with probabilistic LLM prior, versus vanilla PBO | PBO · Idealista18 Madrid (LLM simulation) | pairwise accuracy: 0.539 -> 0.613 | Improved |
| run PBO in the autoencoder latent space with probabilistic LLM prior, versus vanilla PBO | PBO · Idealista18 Madrid (LLM simulation) | NDCG@10: 0.622 -> 0.706 | Improved |
| add autoencoder latent space with static prior to PBO | PBO · Idealista18 Madrid (statistical simulation) | pairwise accuracy: 0.510 -> 0.556 | Improved |
| add autoencoder latent space with static prior to PBO | PBO · Idealista18 Madrid (statistical simulation) | NDCG@10: 0.658 -> 0.584 | Hurt |
| add probabilistic LLM prior instead of direct LLM weight prior (ablation) | PBO+AE · Idealista18 Madrid | pairwise accuracy: 0.488 -> 0.613 | Improved |
| add probabilistic LLM prior instead of static prior (ablation) | PBO+AE · Idealista18 Madrid | pairwise accuracy: 0.605 -> 0.613 | Improved |
| add probabilistic LLM prior instead of static prior (ablation) | PBO+AE · Idealista18 Madrid | NDCG@10: 0.685 -> 0.706 | Improved |
| add LLM-driven direct weight specification instead of static prior (ablation) | PBO+AE · Idealista18 Madrid | pairwise accuracy: 0.605 -> 0.488 | Hurt |
| add probabilistic LLM prior on Munich dataset instead of static prior (ablation) | PBO+AE · Munich rental dataset | pairwise accuracy: 0.492 -> 0.569 | Improved |
| add autoencoder latent space with probabilistic prior on Munich dataset | PBO · Munich rental dataset | pairwise accuracy: 0.544 -> 0.569 | Improved |
| add autoencoder latent space with probabilistic prior with open-source LLM | PBO · Idealista18 Madrid (gpt-oss-120b simulation) | pairwise accuracy: 0.504 -> 0.573 | Improved |
| add LLM warm start to PBO+AE (ablation) | PBO+AE · Idealista18 Madrid (LLM simulation) | pairwise accuracy | Improved |
Closest papers by meaning
- Limited Preference Data? Learning Better Reward Model with Latent Space Synthesis (NeurIPS 2025 · similarity 0.85)
- Preference Diffusion for Recommendation (ICLR 2025 · similarity 0.85)
- Deep Bayesian Active Learning for Preference Modeling in Large Language Models (NeurIPS 2024 · similarity 0.85)
- Preference Learning of Latent Decision Utilities with a Human-like Model of Preferential Choice (NeurIPS 2024 · similarity 0.84)
- MAPLE: A Framework for Active Preference Learning Guided by Large Language Models (AAAI 2025 · similarity 0.84)
- Preference Learning Algorithms Do Not Learn Preference Rankings (NeurIPS 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
