The paper investigates how the ability to synthesize arbitrary queries (membership queries) changes the sample complexity of active learning compared to the traditional pool-based setting. It shows that some hypothesis classes that only achieve polynomial error decay with pool-based queries become exponentially learnable when synthesis is allowed, revealing a significant gap in learning difficulty. The authors propose sufficient conditions, provide examples, and suggest a conjectural framework to identify classes that benefit from synthesized queries.
By Ganghua Wang, Shaddin Dughmi
arXiv:2607. 19199v1 Announce Type: new Abstract: Offline reinforcement learning (RL) aims to learn an effective policy from a static dataset, but its performance is fundamentally limited by dataset coverage.
By Li-Rong Zhou, Qin-Wen Luo, Sheng-Jun Huang
The paper introduces SELECT, an algorithmic framework for satisficing regret minimization in bandit problems, achieving constant expected satisficing regret when a satisficing arm exists. A variant, SELECT‑LITE, further ensures a light‑tailed satisficing regret distribution while maintaining constant expected regret in the realizable case and sub‑linear standard regret otherwise. Experiments on synthetic data and a real‑world dynamic pricing scenario demonstrate the practical effectiveness of both algorithms.
By Qing Feng, Tianyi Ma, Ruihao Zhu
arXiv:2602. 02061v2 Announce Type: replace Abstract: Explosive demands for LLMs often cause user queries to accumulate in server queues, requiring efficient routing (query-LLM matching) and scheduling (query prioritization) mechanisms.
By Seoungbin Bae, Junyoung Son, Dabeen Lee
arXiv:2603. 08001v3 Announce Type: replace Abstract: Maximum inner product search (MIPS) is a crucial subroutine in machine learning, requiring the identification of a vector taken within a database (the keys) that best aligns with a given query.
By Theo X. Olausson, Jo\~ao Monteiro, Michal Klein, Marco Cuturi
arXiv:2603. 08001v2 Announce Type: replace Abstract: Maximum inner product search (MIPS) is a crucial subroutine in machine learning, requiring the identification of a vector taken within a database (the keys) that best aligns with a given query.
By Theo X. Olausson, Jo\~ao Monteiro, Michal Klein, Marco Cuturi
arXiv:2411. 03253v2 Announce Type: replace-cross Abstract: We propose a general framework for end-to-end learning of data structures.
By Omar Salemohamed, Laurent Charlin, Shivam Garg, Vatsal Sharan, Gregory Valiant
arXiv:2307.02719v5 Announce Type: replace
Abstract: Uncertainty sampling is a classical active-learning strategy, yet the statistical objective induced by its query rule is often implicit. We introdu...
By Shang Liu, Xiaocheng Li
arXiv:2510. 21431v2 Announce Type: replace-cross Abstract: We study the combinatorial semi-bandit problem where an agent selects a subset of base arms and receives individual feedback.
By Jung-hun Kim, Milan Vojnovi\'c, Min-hwan Oh
arXiv:2606. 08410v1 Announce Type: cross Abstract: Personalized decision-making in multi-objective bandits requires learning user-specific trade-offs among competing objectives.
By Linfeng Cao, Ming Shi, Ness B. Shroff
arXiv:2509. 05663v4 Announce Type: replace Abstract: Truly unsupervised approaches for time series anomaly detection are rare in the literature.
By Lucas Correia, Jan-Christoph Goos, Thomas B\"ack, Anna V. Kononova
arXiv:2606. 17805v1 Announce Type: new Abstract: Data acquisition is a major bottleneck for learning in real-time streams: analysts must decide on the fly which labels to purchase while respecting a rolling budget.
By Xiwen Huang, Pierre Pinson