This paper studies the problem of quantifying dataset hardness for learned indexes and proposes two scores, conformance and coverage, that together characterize metric quality while capturing a fundamental tradeoff between metric accuracy and applicability.
This work proposes differentially private algorithms for MSD under both cardinality and matroid constraints, achieving nearly optimal utility guarantees and design more efficient algorithms that maintain strong guarantees.
Ron Zadicario, Tova Milo· Proceedings of the VLDB Endo...· 1 citation
LowRankArena is presented, a standardized evaluation platform for SVD-based LLM compression that unifies task versions, uniform-precision compression budgets, comparison regimes, and inference measurements, and provides a reproducible pipeline with over 3 TiB released compressed checkpoints.
Zi-Shan Shao, Li-Xun Zhang, Kang-Ning Cui et al.· 0 citations
A persistent research-practice gap is identified in which algorithmic innovation has outpaced system-level integration, underscoring the infrastructural work still required for production adoption and contributing a comprehensive architectural taxonomy, an evidence-based performance framework, and a prioritized researc...
Manar Abdelhamid, D. Elzanfaly, K. Nagaty et al.· Journal of Big Data· 0 citations
This work argues that benchmark evaluation should also consider the data-centric peak performance frontier, defined by the best statistically supported performance achieved on each dataset, and finds that common aggregation metrics are highly correlated and largely measure consistency and avoiding failures, while being...
Andrej Tschalzev, Stefan Lüdtke, Heiner Stuckenschmidt et al.· 0 citations
This work asks what a compression statistic can justify when deployment cares about the worst supplied group, and treats each statistic as an information interface.
This work implements a suite of algorithms, covering the literature and new methods, both with theoretical guarantees and heuristics, and evaluates their performance on real and synthetic data, both inherently non metric and corrupted, demonstrating that the algorithms' performance is determined predominantly by the ty...
Asaf Etgar, A. Gilbert· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
