SOLAR is a learning-augmented policy for semantic cache replacement that achieves constant competitive ratio 3 and 5-75% gains over FIFO on retrieval workloads.
Bandit Algorithms
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
years
2026 8roles
background 2polarities
background 2representative citing papers
Derives first lower bound on γ_t for mean-based algorithms in unknown-horizon bandit settings, proposes two new algorithms, and shows some are also no-regret.
The paper introduces the Worst-case Marginal Benefit (WMB) criterion for sample-size design in test-and-roll experiments and shows it yields an optimal m approximately equal to N/3 for Bernoulli and Gaussian outcomes.
SAGA introduces workflow-atomic scheduling for compound AI agents, achieving 1.64x lower task completion time and 1.22x better memory utilization than vLLM on a 64-GPU cluster at the cost of 30% lower peak throughput.
Introduces BOBa, a multi-armed bandit method for scalable surrogate optimization that adaptively allocates inference and evaluations to promising partitions of ultra-large chemical libraries.
A classical agent extracts more work from quantum temporal correlations via adaptive strategies bounded by the new Time-Ordered Free Energy, while reinforcement learning achieves polylogarithmic dissipation when learning unknown states.
CSTS learns context-dependent weights for multiple objectives in a multi-objective contextual bandit and outperforms fixed-weight and standard contextual bandit baselines on Swiss public broadcaster programming data.
Bayesian optimization automates the scientific discovery cycle by modeling observations with surrogate models and using acquisition functions to select experiments that balance known information with new exploration.
citing papers explorer
-
SAGA: Workflow-Atomic Scheduling for AI Agent Inference on GPU Clusters
SAGA introduces workflow-atomic scheduling for compound AI agents, achieving 1.64x lower task completion time and 1.22x better memory utilization than vLLM on a 64-GPU cluster at the cost of 30% lower peak throughput.
-
Efficient and Principled Scientific Discovery through Bayesian Optimization: A Tutorial
Bayesian optimization automates the scientific discovery cycle by modeling observations with surrogate models and using acquisition functions to select experiments that balance known information with new exploration.