RHC-UCRL is the first algorithm for safety-constrained RL under explicit adversarial dynamics, providing sub-linear regret and constraint violation guarantees by maintaining optimism over both agent and adversary policies.
hub
Gaussian Process Optimization in the Bandit Setting: No Regret and Experimental Design
26 Pith papers cite this work, alongside 1,053 external citations. Polarity classification is still indexing.
26
Pith papers citing it
1,053
external citations · Pith
abstract
Many applications require optimizing an unknown, noisy function that is expensive to evaluate. We formalize this task as a multi-armed bandit problem, where the payoff function is either sampled from a Gaussian process (GP) or has low RKHS norm. We resolve the important open problem of deriving regret bounds for this setting, which imply novel convergence rates for GP optimization. We analyze GP-UCB, an intuitive upper-confidence based algorithm, and bound its cumulative regret in terms of maximal information gain, establishing a novel connection between GP optimization and experimental design. Moreover, by bounding the latter in terms of operator spectra, we obtain explicit sublinear regret bounds for many commonly used covariance functions. In some important cases, our bounds have surprisingly weak dependence on the dimensionality. In our experiments on real sensor data, GP-UCB compares favorably with other heuristical GP optimization approaches.
hub tools
citation-role summary
background 3
method 1
citation-polarity summary
representative citing papers
STEMGym benchmark demonstrates that perception pipelines dominate dose efficiency in autonomous STEM over navigation methods across 33 agent setups.
Conditional diffusion models trained with BO-aware strategies approximate the optimum distribution, enabling a Diffusion-based Mode Seeking acquisition function with a sub-optimality guarantee that outperforms baselines in experiments.
SILO outperforms five baselines on eight protein fitness landscapes by using trajectory-level imitation on trajectories selected via hierarchical beam search and biological proxy guidance under limited oracle budgets.
AB-SID-iVAR enables Gaussian process active learning for self-induced Boltzmann distributions by closed-form approximation of the target, with high-probability error vanishing guarantees and empirical gains on PES and drug discovery tasks.
FORGE reformulates molecular optimization as context-aware fragment ranking and replacement using mined low-to-high edit pairs, outperforming larger language models and graph methods on standard benchmarks.
Probability-of-Hit acquisition function ranks perturbation candidates by posterior probability of threshold exceedance, with asymptotic optimality proof and up to 6.4% gains on real immunology data.
A myopic MINMPC framework learns a value function offline via inverse optimization from expert data, allowing short horizons with near-optimal performance and strict integer feasibility online for hybrid systems.
Spectral bandits achieve scalable regret in graph-structured recommendation by using an effective dimension to learn good policies from few node evaluations.
Characterizes the optimal asymptotic competitive ratio for parametric prophet inequalities and proposes an online confidence-based DP policy achieving it without offline samples.
CRAFT is a Pareto-front prompt optimizer that allocates scarce LLM validation calls to candidates near the current front using accuracy- and cost-oriented generators plus NSGA-II retention.
LGBO integrates LLM semantic preferences continuously into Bayesian optimization iterations, with a theoretical worst-case guarantee and empirical gains including 90% of best value in 6 iterations on a wet-lab battery task.
NEON provides uncertainty-aware operator learning for composite Bayesian optimization in function spaces using a single network, achieving claimed SOTA with orders of magnitude fewer parameters than ensembles.
A mixed-variable Bayesian optimization framework based on latent variable Gaussian processes is developed and demonstrated on optimizing composition and morphology for insulating polymer nanocomposites, with an extension to multi-objective Pareto optimization.
ADKO is a decentralized framework where agents share compact GP-derived tokens and LM insights to achieve collaborative Bayesian optimization with a decomposed regret bound that includes compression and approximation losses.
Decoupled PFNs use controllable synthetic priors to train separate latent-signal and noise heads, making epistemic-aleatoric decomposition identifiable and improving acquisition in noisy settings.
Lightweight numerical bandits on text embeddings match or exceed LLM accuracy in contextual bandits at a fraction of the cost, with an embedding-based diagnostic to choose between them.
MAPFZ extends classical MAPF to non-unit integer costs on graphs with finite states, solved efficiently by CBS-NIC and Bayesian-optimized discretization, outperforming prior methods on benchmarks.
Proposes projected quantum kernels with misspecified GP bandit algorithms and regret bounds to trade off expressivity against learnability in quantum kernel optimization.
P-K-GCN integrates continuous spline GCN, Koopman linearization, and physics augmentation for spatiotemporal super-resolution on irregular geometries, claiming theoretical error reduction via Rademacher complexity bounds and superior accuracy on cardiac electrodynamics.
The paper derives provably tighter instantaneous regret bounds for GP-UCB and proposes (ε,δ)-optimal stopping criteria for Bayesian optimization based on those bounds.
A restarting-based nonparametric online learning method for dynamic pricing with one-point revenue feedback that achieves regret bounds scaling with time horizon and total market variation.
A dual-ranking strategy improves offline data-driven multi-objective optimization by prioritizing solutions that score well on both predicted performance and low uncertainty across different surrogate models.
Formalizes budget-constrained posterior dialog orchestration as CABO and evaluates the approach on simulated and proprietary conversational datasets.
citing papers explorer
-
Optimistic Policy Learning under Pessimistic Adversaries with Regret and Violation Guarantees
RHC-UCRL is the first algorithm for safety-constrained RL under explicit adversarial dynamics, providing sub-linear regret and constraint violation guarantees by maintaining optimism over both agent and adversary policies.
-
STEMGym: Benchmarking Sequential Decision-Making under Dose Budgets in Autonomous Electron Microscopy
STEMGym benchmark demonstrates that perception pipelines dominate dose efficiency in autonomous STEM over navigation methods across 33 agent setups.
-
Improving Bayesian Optimization via Training-Aware Conditional Diffusion Models
Conditional diffusion models trained with BO-aware strategies approximate the optimum distribution, enabling a Diffusion-based Mode Seeking acquisition function with a sub-optimality guarantee that outperforms baselines in experiments.
-
Self-Improvement Imitation with Biologically Guided Search for Protein Design Under Oracle Budgets
SILO outperforms five baselines on eight protein fitness landscapes by using trajectory-level imitation on trajectories selected via hierarchical beam search and biological proxy guidance under limited oracle budgets.
-
Active Learning for Gaussian Process Regression Under Self-Induced Boltzmann Weights
AB-SID-iVAR enables Gaussian process active learning for self-induced Boltzmann distributions by closed-form approximation of the target, with high-probability error vanishing guarantees and empirical gains on PES and drug discovery tasks.
-
FORGE: Fragment-Oriented Ranking and Generation for Context-Aware Molecular Optimization
FORGE reformulates molecular optimization as context-aware fragment ranking and replacement using mined low-to-high edit pairs, outperforming larger language models and graph methods on standard benchmarks.
-
Many Needles in a Haystack: Active Hit Discovery for Perturbation Experiments
Probability-of-Hit acquisition function ranks perturbation candidates by posterior probability of threshold exceedance, with asymptotic optimality proof and up to 6.4% gains on real immunology data.
-
Learning myopic mixed-integer nonlinear model predictive control from expert demonstrations
A myopic MINMPC framework learns a value function offline via inverse optimization from expert data, allowing short horizons with near-optimal performance and strict integer feasibility online for hybrid systems.
-
Spectral bandits
Spectral bandits achieve scalable regret in graph-structured recommendation by using an effective dimension to learn good policies from few node evaluations.
-
Asymptotically Optimal Learning for Parametric Prophet Inequalities
Characterizes the optimal asymptotic competitive ratio for parametric prophet inequalities and proposes an online confidence-based DP policy achieving it without offline samples.
-
CRAFT: Cost-aware Refinement And Front-aware Tuning of Prompts
CRAFT is a Pareto-front prompt optimizer that allocates scarce LLM validation calls to candidates near the current front using accuracy- and cost-oriented generators plus NSGA-II retention.
-
Unleashing LLMs in Bayesian Optimization: Preference-Guided Framework for Scientific Discovery
LGBO integrates LLM semantic preferences continuously into Bayesian optimization iterations, with a theoretical worst-case guarantee and empirical gains including 90% of best value in 6 iterations on a wet-lab battery task.
-
Composite Bayesian Optimization In Function Spaces Using NEON -- Neural Epistemic Operator Networks
NEON provides uncertainty-aware operator learning for composite Bayesian optimization in function spaces using a single network, achieving claimed SOTA with orders of magnitude fewer parameters than ensembles.
-
Data-Centric Mixed-Variable Bayesian Optimization For Materials Design
A mixed-variable Bayesian optimization framework based on latent variable Gaussian processes is developed and demonstrated on optimizing composition and morphology for insulating polymer nanocomposites, with an extension to multi-objective Pareto optimization.
-
ADKO: Agentic Decentralized Knowledge Optimization
ADKO is a decentralized framework where agents share compact GP-derived tokens and LM insights to achieve collaborative Bayesian optimization with a decomposed regret bound that includes compression and approximation losses.
-
Decoupled PFNs: Identifiable Epistemic-Aleatoric Decomposition via Structured Synthetic Priors
Decoupled PFNs use controllable synthetic priors to train separate latent-signal and noise heads, making epistemic-aleatoric decomposition identifiable and improving acquisition in noisy settings.
-
When Do We Need LLMs? A Diagnostic for Language-Driven Bandits
Lightweight numerical bandits on text embeddings match or exceed LLM accuracy in contextual bandits at a fraction of the cost, with an embedding-based diagnostic to choose between them.
-
Multi-Agent Pathfinding with Non-Unit Integer Edge Costs via Enhanced Conflict-Based Search and Graph Discretization
MAPFZ extends classical MAPF to non-unit integer costs on graphs with finite states, solved efficiently by CBS-NIC and Bayesian-optimized discretization, outperforming prior methods on benchmarks.
-
Balancing Expressivity and Learnability in Quantum Kernel Bandit Optimization
Proposes projected quantum kernels with misspecified GP bandit algorithms and regret bounds to trade off expressivity against learnability in quantum kernel optimization.
-
P-K-GCN: Physics-augmented Koopman-enhanced Graph Convolutional Network for Deep Spatiotemporal Super-resolution
P-K-GCN integrates continuous spline GCN, Koopman linearization, and physics augmentation for spatiotemporal super-resolution on irregular geometries, claiming theoretical error reduction via Rademacher complexity bounds and superior accuracy on cardiac electrodynamics.
-
Regret-Based $(\epsilon,\delta)$-optimal Stopping Criteria for Bayesian Optimization
The paper derives provably tighter instantaneous regret bounds for GP-UCB and proposes (ε,δ)-optimal stopping criteria for Bayesian optimization based on those bounds.
-
Nonparametric Learning and Earning with One-Point Feedback under Nonstationarity
A restarting-based nonparametric online learning method for dynamic pricing with one-point revenue feedback that achieves regret bounds scaling with time horizon and total market variation.
-
Uncertainty-Aware Offline Data-Driven Multi-Objective Optimization
A dual-ranking strategy improves offline data-driven multi-objective optimization by prioritizing solutions that score well on both predicted performance and low uncertainty across different surrogate models.
-
A Bandit Approach to Posterior Dialog Orchestration Under a Budget
Formalizes budget-constrained posterior dialog orchestration as CABO and evaluates the approach on simulated and proprietary conversational datasets.
-
Laser-Enhanced Contact Optimization in Silicon Photovoltaics: Mechanisms, Reliability, and Predictive Process Design
A review of LECO in silicon photovoltaics that frames it as a multiphysics process and outlines a predictive workflow using regime maps and reduced state metrics for stable contact optimization.
- Bayesian Optimization of Crossbar-Based Compute-In-Memory System Design for Efficient DNN Inference