New RLHF framework handles general preference oracles without reward functions.
problem Handling general preference oracles without assuming a reward function.
method Developed a minimax game between two LLMs for RLHF under a general preference oracle, focusing on KL-regularized preference.
result Proposed algorithms for efficient offline and online RLHF learning.
Derives a new objective to learn from human preferences without approximations.
problem Learning from human preferences through RLHF relies on approximations that can lead to pitfalls.
method Derives a new general objective ΨPO that bypasses both approximations. result Demonstrates the superiority of the new objective to Direct Preference Optimisation (DPO) empirically.
Bayesian optimization agent learns user preferences from pairwise comparisons.
problem Learning user preferences from unknown and infinite choices.
method Sequential Bayesian optimization with pairwise comparisons.
result Optimal agent strategy minimizes remaining system uncertainty.
Tutorials on preference learning with Gaussian Processes.
problem Understanding individual preferences and choices for efficient and personalized applications.
method Presentation of a comprehensive framework for preference learning with Gaussian Processes, incorporating rationality principles.
result Construction of preference learning models that encompass various utility models and scenarios.
Model learns metrics and preferences from user comparisons.
problem Simultaneous metric and preference learning from user comparisons.
method Jointly learns a metric and latent ideal points for each user.
result Model captures individual preferences and learns metrics efficiently.
SLHF uses sequential game theory to optimize preferences from human feedback.
problem Optimizing preferences from human feedback in sequential settings.
method SLHF frames the problem as a sequential-move game between Leader and Follower, decomposing the optimization into refinement and adversarial optimization.
result SLHF achieves strong alignment across diverse preference datasets and scales to large models.
Paper explores limits and possibilities of aligning LLMs with human preferences.
problem Aligning LLMs with diverse human preferences to ensure fairness and informed outcomes.
method Analysis of probabilistic representation of human preferences and preservation of diverse preferences.
result LLMs can't fully align with human preferences using reward-based approaches due to Condorcet cycles, but mixed strategies are statistically possible.
Paper shows RLHF can be solved similarly to standard RL.
problem Difficulty of RLHF compared to standard RL.
method Reduction to reward-based RL techniques.
result RLHF can be solved using existing algorithms for reward-based RL.
New method accounts for hidden context in preference learning for RLHF models.
problem Incomplete data with hidden context affects RLHF model outcomes.
method Distributional Preference Learning (DPL) methods estimate hidden context distributions.
result DPL methods reduce RLHF vulnerabilities by accounting for hidden context.
Enhances preference learning by incorporating response times into binary choices.
problem Limited information from binary choices about preference strength.
method Combines choices and response times using the EZ diffusion model.
result Response times improve utility estimation for strong preferences.
Study on learning strategies in matching markets with uncertain preferences.
problem Decision-making in scarcity of shared resources with unknown agent preferences.
method Representation of preferences in a reproducing kernel Hilbert space, learning algorithm for uncertainty.
result Optimal strategies derived to maximize agents' expected payoffs, with stability and fairness properties.
DOPL learns from preference feedback to solve RMAB problems.
problem Learning optimal decisions in RMAB with limited reward information.
method Direct online preference learning (DOPL) for Pref-RMAB.
result DOPL achieves sublinear regret for RMAB with preference feedback.
In preference-based reinforcement learning (RL), an agent interacts with the environment while receiving preferences instead of absolute feedback. While there is increasing research activity in preference-based RL, the design of formal frameworks that admit tractable theoretical analysis remains an open challenge. Buil…
This paper improves sample efficiency for off-policy evaluation with preference data.
problem Improving sample efficiency for off-policy evaluation with preference data.
method Using a deep neural network to learn the value function and leveraging manifold structure.
result Established a provably efficient guarantee for off-policy evaluation with RLHF.
IDT learns human preferences from uncertain decisions, even when humans are suboptimal.
problem Learning human preferences from uncertain and suboptimal decisions.
method Inverse decision theory (IDT) framework, statistical analysis of IDT, characterizing sample complexity.
result Learning preferences is easier when decisions are more uncertain, even if humans are suboptimal.
The paper proposes a method to infer multi-objective rewards from preferences.
problem Modeling preferences based on multiple, often competing objectives.
method Modeling priorities lexicographically and inferring multi-objective rewards from observed preferences.
result Lexicographically-ordered rewards provide a better understanding of preferences and improve policies.
This work proves win rate is key to understanding preference learning.
problem Understanding preference learning from generative models.
method Analyzing preference learning methods as win rate optimization or non-WRO.
result Proves win rate is the only evaluation respecting preferences and prevalences.
Meta-algorithm for efficient reinforcement learning from human preferences.
problem Learning from human preference comparisons in Markov decision processes.
method Randomized exploration and experimental design for batch comparison queries.
result Meta-algorithm achieves both regret and last-iterate guarantees with minimal preference queries.
We state the problem of inverse reinforcement learning in terms of preference elicitation, resulting in a principled (Bayesian) statistical formulation. This generalises previous work on Bayesian inverse reinforcement learning and allows us to obtain a posterior distribution on the agent's preferences, policy and optio…
Improved DPO framework penalizes preference uncertainty to avoid overoptimization.
problem Aligning LLMs to human preferences is challenging due to varied, context-dependent, and ambiguous preferences.
method Developed a pessimistic framework for DPO by introducing preference uncertainty penalization schemes.
result Improved overall performance and better completions on high-uncertainty responses compared to vanilla DPO.
A new approach to fine-tuning LLMs with human feedback.
problem Inability of current reward models to fully represent human preferences.
method Introducing NLHF, a new pipeline for LLM fine-tuning using pairwise human feedback.
result NLHF produces a sequence of policies converging to the regularized Nash equilibrium.
Optimizes crowdsourced preference-based subjective evaluation with online learning.
problem Large-scale evaluation of generative media using crowdsourcing due to combinatorial explosion.
method Automatic optimization of pair combination selections and evaluation volumes with online learning.
result Optimizes evaluation by reducing pair combinations and allocating optimal evaluation volumes.
Study preference-based reinforcement learning in episodic kernel MDPs.
problem Learning from episodic human preferences in reinforcement learning.
method Developed preference-based value estimation and confidence sets for kernel-based MDPs.
result Proved high-probability regret bounds that converge to optimal policy value.
Efficiently learns reward functions with fewer queries and shorter computation times.
problem Expensive data generation and labeling in robot learning.
method Batch active preference-based learning methods using determinantal point processes (DPP) and heuristic alternatives.
result Our batch active learning algorithm requires only a few queries and computes them in a short amount of time.
SafeMIL learns safer policies by avoiding risky behavior from non-preferred trajectories.
problem Learning safe imitation policies from non-preferred trajectories in risky environments.
method SafeMIL uses Multiple Instance Learning to learn a cost function from non-preferred trajectories.
result SafeMIL learns a safer policy that avoids non-preferred behaviors without sacrificing reward performance.
This paper learns user preferences from comparisons using Mahalanobis metrics.
problem Learning user preferences from paired comparisons with unknown metric.
method Estimates user ideal point and Mahalanobis metric from comparisons.
result Effectiveness of the algorithm demonstrated on synthetic and real-world data.
New algorithms learn stable matchings from uncertain user preferences.
problem Learning stable matchings from uncertain user preferences.
method Stochastic multi-armed bandit problem, incentive-aware learning objective, primal-dual formulation.
result Near-optimal regret bounds for learning stable matchings.
Study tackles RLHF with diverse human feedback, showing limitations and proposing a meta-learning approach.
problem Traditional RLHF fails to balance diverse human preferences.
method Integrates meta-learning and multiple social welfare functions to optimize diverse preferences.
result Establishes sample complexity bounds for optimizing diverse social welfare functions.
Inverse reinforcement learning (IRL) enables an agent to learn complex behavior by observing demonstrations from a (near-)optimal policy. The typical assumption is that the learner's goal is to match the teacher's demonstrated behavior. In this paper, we consider the setting where the learner has its own preferences th…
Enhances math problem-solving models with multi-turn preference learning.
problem Improving mathematical problem-solving capabilities of large language models.
method Introduces a multi-turn direct preference learning framework for tool-integrated mathematical reasoning tasks.
result Significant performance improvements in model accuracy on math datasets.
The paper enhances preference learning by incorporating response time data.
problem Lack of temporal information in user decision-making for reward model learning.
method Integrates response time alongside binary choice data using the EZ model and Neyman-orthogonal loss functions.
result Response time-augmented approach reduces error rates from exponential to polynomial scaling, improving sample efficiency.
New algorithm for learning preferences in decentralized matching markets reduces regret to logarithmic levels.
problem Learning preferences in decentralized matching markets without direct communication.
method Introduces a new algorithm for two-sided matching markets with competition.
result The algorithm achieves logarithmic stable regret in shared preferences and quadratic regret in general preferences.
Bayesian optimization learns DM preferences for multi-outcome experiments.
problem Optimizing expensive experiments with unknown utility functions and multiple outcomes.
method Alternates preference learning and Bayesian optimization, using pairwise comparisons.
result Preference exploration strategies improve Bayesian optimization performance.
Paper investigates monotonicity issues in AI preference learning.
problem AI models may violate monotonicity when learning preferences.
method Investigates root causes of non-monotonicity in comparison-based preference learning.
result Proves local pairwise monotonicity under mild assumptions.
Active learning framework for optimizing human preferences in reinforcement learning.
problem Selecting most informative feedback for training models of human preferences.
method Proposes an active learning framework to collect preferential feedback online or offline.
result Errors in DPO logit estimates diminish with more feedback.
ACOL learns constraints from human preferences in driving simulations.
problem Learning constraints from human preferences in driving simulations.
method Adaptive Constraint Learning (ACOL) algorithm for constrained linear best-arm identification.
result ACOL's sample complexity matches worst-case lower bound and is significantly tighter in the average case.
The paper tackles statistical and computational challenges in learning correlated reward models.
problem The Independence of Irrelevant Alternatives (IIA) assumption collapses human preferences into a universal utility function, leading to coarse approximations.
method The paper investigates the statistical and computational challenges of learning a correlated probit model using best-of-three preference data.
result Best-of-three preference data overcomes the limitations of pairwise preference data, allowing for more fine-grained modeling of human preferences.
Robot motions in the presence of humans should not only be feasible and safe, but also conform to human preferences. This, however, requires user feedback on the robot's behavior. In this work, we propose a novel approach to leverage the user's brain signals as a feedback modality in order to decode the judgment of rob…
Best-of-N sampling reveals reward targets from preference data, influencing N and base distribution choices.
problem Understanding reward extraction from Best-of-N preference data and optimal N and base distribution choices.
method Specialized analysis of preference data via induced conditional distribution, deriving reward targets and design principles.
result Reward targets are explicit functions of N and base distribution, and bounded-class minimizers approach these targets as N grows.
The paper explores game-theoretic alignment of LLMs with human preferences, finding limitations and conditions.
problem Aligning LLMs with human preferences using game theory.
method Systematic study of payoff choices in a two-player zero-sum game for desirable alignment properties.
result Impossibility of preference matching in game-theoretic LLM alignment under standard assumptions.
Paper improves parameter estimation of continuous distributions using preference feedback.
problem Improving parameter estimation of continuous distributions.
method Preference-based M-estimators and deterministic preferences.
result Preference-based estimators achieve an estimation error scaling of O(1/n), significantly faster than sample-only methods.
CnGAN generates synthetic user preferences for non-overlapped users in cross-network recommender systems.
problem Cross-network recommender solutions ignore non-overlapped users, limiting their applicability.
method Multi-task learning, encoder-GAN architecture, user-based pairwise loss function.
result Generated user preferences improve recommendations for non-overlapped users, achieving superior performance.
OSIL learns safe policies from unsafe demonstrations.
problem Offline safe imitation learning with implicit safety.
method Formulates CMDP, infers safety from non-preferred trajectories, learns cost model.
result OSIL learns safer policies without degrading reward performance.
The paper proposes a method to learn and leverage contextual preference distributions for better decision-making.
problem Heterogeneous and context-dependent human preferences in decision-making problems.
method A sequential learning-and-optimization pipeline using a bounded-variance score function gradient estimator to train a predictive model mapping contextual features to preference distributions.
result The approach reduces average post-decision surprise by up to 25 times compared to risk-averse baselines in a ridesharing environment.
This work compares human feedback methods for reward learning in bandits.
problem Understanding how human feedback affects the performance of reward learning methods.
method Theoretical comparison of human feedback approaches in offline contextual bandits.
result Human bias and uncertainty in feedback modeling impact the theoretical guarantees of reward learning methods.
Bayesian optimization with preference learning identifies preferred solutions in multi-objective problems.
problem Optimizing multiple criteria with decision maker preferences in expensive functions.
method Bayesian optimization with interactive preference learning and active acquisition function.
result Identifies the most preferred solution with reduced interaction cost.
Thermal preferences vary from person to person and may change over time. The main objective of this paper is to sequentially pose intelligent queries to occupants in order to optimally learn the indoor air temperature values which maximize their satisfaction. Our central hypothesis is that an occupant's preference rela…
We consider black-box global optimization of time-consuming-to-evaluate functions on behalf of a decision-maker (DM) whose preferences must be learned. Each feasible design is associated with a time-consuming-to-evaluate vector of attributes and each vector of attributes is assigned a utility by the DM's utility functi…