Pareto Optimization

Follow this topic Ask about this topic

Pareto optimization studies how to find and represent solutions when improving one objective can worsen another, such as accuracy versus cost or fairness versus utility. Methods include weighted and Chebyshev scalarization, evolutionary and gradient-based front learning, Bayesian search, and preference-guided navigation; applications range from control and routing to machine learning and resource allocation. The goal is often to give decision-makers a useful set of trade-offs rather than one compromise solution. A central challenge is obtaining broad, reliable coverage of the Pareto front under computational, data, and feasibility constraints, since scalarization can miss non-convex regions and computing the full front may be expensive.

Latest papers

40 papers3 papers in the last four weeks, against 2 the four weeks before. 0.0% of all new papers.

Beyond Scalar Leaderboards: Adaptive Sampling for Pareto Frontier Identification in Multi-Objective LLM Arenas
Oct 20, 2026Liang Twist Shan, Jiahao Ai, Zhimei RenLLM Evaluation Adaptive Sampling
Department of Statistics, University of Wisconsin-Madison · Department of Statistics and Data Science, University of Pennsylvania
In-Context Preference-Conditioned Pareto Set Learning from Limited Offline Data
Oct 15, 2026Yiming Yao, Liang Zhao, Qingfu ZhangMulti-Objective Optimization Pareto Optimization
City University of Hong Kong
Learning Pareto Stationary Fronts via Single-Pass Backpropagation
Oct 5, 2026Elina Rojin Celik, Marcos Medeiros Raimundo, Isabel ValeraDeep Learning Optimization Multi-Objective Optimization
Department of Computer Science, Saarland University, Germany · Instituto de Computação, Universidade Estadual de Campinas, Brazil
Exact Regret Frontiers and Externality Scheduling in Centralized Serial-Dictatorship Bandits
Sep 17, 2026Lishang Xu, Guodong Ma, Pengcheng Weng +1Multi-Armed Bandits Regret Minimization
University of Bern · EPFL
ParetoTransport: Generative Optimization by Mass Transport Toward The Pareto Front
Sep 7, 2026Stephanie Holly, Sepp Hochreiter, Werner ZellingerMulti-Objective Optimization Pareto Optimization
LIT AI Lab and Institute for Machine Learning JKU Linz, Austria
Cost-Aware Multi-Objective Bandits: Theory and Application to Budgeted LLM Configuration Evaluation
Aug 5, 2026Bo Xue, Zhi Hong, Jiayi Li +3Multi-Armed Bandits LLM Evaluation
City University of Hong Kong · South China University of Technology · Zhejiang University
Top- $k$ Pareto Bandits: Hypervolume Regret for Multi-Objective Slate Selection
Jul 28, 2026Nicolas Gutowski, Fabien Chhel, Alexandre Letard +1Multi-Armed Bandits Combinatorial Multi-Armed Bandits
Université d’Angers, LERIA, SFR MATHSTIC, Angers, France · ESEO, ERIS, Angers, France · AlphaEdge, Saint-Barthélémy-d’Anjou, France +1
Fair Division with Strictly Increasing Valuations: A Tight Threshold for Two-Agent EF1 and PO
Jul 25, 2026Nicholas TehSocial Choice Theory Fair Resource Allocation
University of Oxford, UK
Representing the Non-dominated Set of Multi-objective Network Problems by Supported Non-dominated Points
Jul 13, 2026David Könen, Lara Löhken, Michael StiglmayrCombinatorial Optimization Multi-Objective Optimization
University of Wuppertal, School of Mathematics and Natural Sciences, Optimization Group, Gaußstraße 20, 42103 Wuppertal, Germany · Corresponding author
Deterministic Pareto-Optimal Policy Synthesis for Multi-Objective Reinforcement Learning
Jun 24, 2026Aniruddha Joshi, Niklas Lauffer, Sanjit SeshiaMulti-Objective Reinforcement Learning Multi-Objective Optimization
University of California Berkeley, USA.
Selective Ensemble Based on Preference-Directed Multi-Objective Bandits
Jun 20, 2026Lanjihong Ma, Zhen-Yu Zhang, Masashi Sugiyama +1Multi-Armed Bandits Ensemble Learning
Department of Computer Science, Zhejiang Gongshang University, China · Center for Advanced Intelligence Project, RIKEN, Japan · Graduate School of Frontier Sciences, The University of Tokyo, Japan +1
Simultaneously Efficient Allocation of Indivisible Items Across Multiple Dimensions
Jun 19, 2026Yasushi Kawase, Bodhayan Roy, Mohammad Azharuddin SanpuiCombinatorial Optimization Approximation Algorithms
Chuo University, Tokyo, Japan · Indian Institute of Technology Kharagpur, Kharagpur, India
Interactive Pareto navigation for deep multi-task learning
Jun 17, 2026Augustina C. Amakor, Konstantin Sonntag, Sebastian PeitzMulti-Task Learning Pareto Optimization
Department of Computer Science, TU Dortmund, Dortmund, Germany · Lamarr Institute for Machine Learning and Artificial Intelligence
Bayesian Anytime Pareto Set Identification for Multi-Objective Multi-Armed Bandits
Jun 17, 2026Lennert Saerens, Bram Silue, Eleni Litsa +2Thompson Sampling Multi-Armed Bandits
Artificial Intelligence Lab, Departement of Computer Science, Vrije Universiteit Brussel, Brussels, Belgium · imec, Leuven, Belgium
Towards Pareto-Optimal Tool-Integrated Agents with Pareto Ranking Policy Optimization
Jun 15, 2026Junyi Li, Xiaowei Qian, Yingyi Zhang +6Multi-Objective Reinforcement Learning Multi-Objective Language Model Alignment
Department of Data Science, City University of Hong Kong, Hong Kong, China · Zhejiang University of Technology, Hangzhou, China · Huawei Technologies Noah’s Ark Lab, Hong Kong, China
Diversity-Driven Offline Multi-Objective Optimization via Nested Pareto Set Learning
Jun 13, 2026Yiyi Zhu, Yaolin Wen, Xiang Xia +6Surrogate-Assisted Optimization Multi-Objective Optimization
Shanghai Institute of AI for Education, and School of Computer Science and Technology, East China Normal University, Shanghai, China · Ant Group, Hangzhou, China · Shanghai Innovation Institute, Shanghai, China
Post-AGI Economies: Superposition and the Second Fundamental Theorem of Welfare Economics
Jun 6, 2026Elija PerrierAI Agent Governance Pareto Optimization
Centre for Quantum Software & Information, University of Technology Sydney, Sydney, Australia
SURF: Steering the Scalarization Weight to Uniformly Traverse the Pareto Front
May 20, 2026Liuyuan Jiang, Chentong Huang, Lisha ChenAdaptive Sampling Multi-Objective Optimization
Department of Electrical and Computer Engineering University of Rochester, Rochester, NY 14627
MOCHA: Multi-Objective Chebyshev Annealing for Agent Skill Optimization
May 19, 2026Md Mehrab Tanjim, Jayakumar Subramanian, Branislav Kveton +6LLM Agent Skill Learning Multi-Objective Optimization
Adobe Research
A Two-Phase Adaptive Balanced Penalty Method for Controllable Pareto Front Learning under Split Feasibility Conditions
May 19, 2026Nguyen Viet Hoang, Dung D. Le, Tran Ngoc ThangHypernetworks Multi-Objective Optimization
Nonsmooth Set-Gradient Ascent to the Pareto Front via Layered Hypervolume and Magnitude Indicators
May 13, 2026Michael T. M. EmmerichMulti-Objective Optimization Pareto Optimization
Faculty of Information Technology, University of Jyväskylä, Finland
Large Language Models as Amortized Pareto-Front Generators for Constrained Bi-Objective Convex Optimization
May 12, 2026Peipei Xu, SiYuan Ma, Yaohua Liu +4Large Language Model-Guided Optimization Constrained Optimization
University of Shanghai for Science and Technology · The Hong Kong University of Science and Technology
Fairness vs Performance: Characterizing the Pareto Frontier of Algorithmic Decision Systems
May 11, 2026Mieke Wilms, Christoph HeitzAlgorithmic Fairness Group Fairness
Zurich University of Applied Sciences / University of Zurich, Switzerland · Zurich University of Applied Sciences, Switzerland
From Feasible to Practical: Pareto-Optimal Synthesis Planning
May 8, 2026Friedrich Hastedt, Dongda Zhang, Antonio del Rio ChanonaMulti-Objective Optimization Pareto Optimization
Department of Chemical Engineering, Imperial College London, UK · Department of Chemical Engineering, University of Manchester, UK
CoAction: Cross-task Correlation-aware Pareto Set Learning
May 3, 2026Xinyue Chen, Yingxuan Liang, Yiqin Huang +3Multi-Task Learning Multi-Task Optimization
Guangdong University of Technology, Guangzhou, China · Xiamen University, Xiamen, China
POCA: Pareto-Optimal Curriculum Alignment for Visual Text Generation
Apr 27, 2026Yaohou Fan, Qingzhong Wang, Yongsong Huang +3Multi-Objective Reinforcement Learning RL for Image Generation
Tohoku University · Amazon Web Services
The Magnitude of Dominated Sets: A Pareto Compliant Indicator Grounded in Metric Geometry
Apr 20, 2026Michael T. M. EmmerichMulti-Objective Optimization Pareto Optimization
Faculty of Information Technology, University of Jyväskylä, Finland
Skyline-First Traversal as a Control Mechanism for Multi-Criteria Graph Search
Apr 14, 2026Nicolas TachenyPareto Optimization Multi-Objective Optimization
Ni2 Innovation Lab, Liège, Belgium · University of Mons (UMONS), Belgium
Beyond Compromise: Pareto-Lenient Consensus for Efficient Multi-Preference LLM Alignment
Apr 7, 2026Renxuan Tan, Rongpeng Li, Zhifeng Zhao +1Preference Alignment Multi-Objective Language Model Alignment
College of Information Science and Electronic Engineering, Zhejiang University · Zhejiang Lab, Hangzhou, China · Macau University of Science and Technology, Macau, China
Many Preferences, Few Policies: Compact Portfolios for Multi-Objective LLM Alignment
Apr 5, 2026Cheol Woo Kim, Jai Moondra, Roozbeh Nahavandi +3Multi-Objective Language Model Alignment Personalized Language Model Alignment
Carnegie Mellon University · Massachusetts Institute of Technology

Oct 20, 2026J/K move · Enter open · S save

Beyond Scalar Leaderboards: Adaptive Sampling for Pareto Frontier Identification in Multi-Objective LLM Arenas

Liang Twist Shan, Jiahao Ai, Zhimei Ren

Department of Statistics, University of Wisconsin-Madison · Department of Statistics and Data Science, University of Pennsylvania

LLM Evaluation Adaptive Sampling Pareto Optimization

Open paper AskPDF arXiv

The evaluation of large language models (LLMs) is inherently multi-objective, as different models may offer different tradeoffs across performance dimensions such as math, code, and writing. Because such evaluations are often based on costly and time-consuming human feedback, it is important to draw reliable conclusions using as few comparisons as possible. In this paper, we present Vector-Borda Empirical-Gap Elimination (VB-EGE), a fixed-confidence adaptive sampling algorithm for efficiently identifying models on the Pareto frontier in arena-style evaluation, where each comparison evaluates a pair of models on a prompt with respect to a selected performance dimension. VB-EGE progressively eliminates models from an active set once sufficient statistical evidence has been accumulated to certify their Pareto status. Under a multi-objective Bradley-Terry model, we show that VB-EGE returns the correct Pareto frontier with high probability and derive an instance-dependent upper bound on its sample complexity. We further establish a worst-case lower bound showing that VB-EGE is minimax-optimal up to logarithmic factors. Simulations and experiments on three real-world Arena datasets (Arena, SearchArena, and VisionArena) demonstrate VB-EGE's empirical effectiveness.