跳到正文
研踪PaperTrail

论文库

检索计算机领域论文，按研究方向、作者或关键词筛选。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+0新增论文
+0其中 AI 相关
+2已有论文修订
+100新增 AI 导读

arXiv 最近完成扫描：2026-11-15 00:32 · 全站最近入库：2026-11-14 15:09各信源进度 →

统计口径与数据新鲜度

统计 2026-11-14 01:11 — 2026-11-15 01:11（北京时间） · 点击数字查看对应简报；AI 相关包含在新增论文内。

最近完成检查：2026-11-15 00:32最近发现新增版本：2026-11-14 15:09库内论文最新更新时间：2026-11-13 02:59

每分钟刷新已保存的数据。空检查不代表采集异常；新收录可能包含历史论文。AI 导读按新增版本计数。

23,348 已收录论文19,580 AI 相关6,729 有修订112 HF 社区发现

浏览论文库

研究日报 / 周报 →按信源浏览 →

时间范围跟随排序：版本更新、本站收录或首次发表。最新收录可能包含历史论文。部分信源异常，查看状态 →

找到 3 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-15 00:30（北京时间）
新论文 · v1arXiv:2611.15975

Towards Unified File Management for Data Lakehouse Systems

Keonwoo Oh, Akshay Mehra, Joshua Kimball, Amol Deshpande

原文摘要

Data lakehouse systems have garnered widespread adoption, yet their metadata management systems face a fundamental architectural dilemma. File-based OTFs (Open Table Formats) scale well, but suffer from metadata bloat, high commit latencies, the absence of multi-table transactions, and unsafe garbage collection. On the other hand, catalogs relying entirely on a transactional database incur severe scalability bottlenecks due to analytic file-level metadata. To reconcile this tension, we present Raven, an operational catalog that stores transactional metadata within a database while offloading analytical metadata to underlying OTF structures. Based on this hybrid architecture, Raven introduces an extended Merge-on-Read (MoR) mechanism to reduce commit latency and mitigate metadata bloat, alongside multi-table transactions under snapshot isolation. Furthermore, Raven unifies core data lakehouse semantics including versioning and reference tracking under an extensible ER (Entity-Relationship) metadata model, enabling safe global garbage collection across diverse types of assets. Empirical evaluations demonstrate that Raven substantially reduces commit latency and metadata bloat when integrated with Apache Iceberg and Delta Lake, maintains a catalog storage footprint orders of magnitude smaller than database-only systems like DuckLake, and achieves scalable garbage collection with minimal disruption.

cs.DB

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2606.18209

Rethinking Dataset Distillation for Classification: Do Distilled Sets Outperform Coresets?

Trisha Mittal, Akshay Mehra, Joshua Kimball

原文摘要

Dataset distillation (DD) has emerged as a prominent approach in data centric machine learning, aiming to synthesize compact training sets for efficient training by compressing the information in large datasets into a small number of synthetic samples. However, DD methods are often evaluated under inconsistent evaluation protocols, ranging from standard ERM to single/multi-teacher supervision, making it difficult to isolate the effectiveness of distilled data from evaluation. Moreover, many prior methods claim that DD outperforms data pruning approaches such as coreset selection (CS), based on the assumption that restricting condensed datasets to subsets of real samples fundamentally limits their expressiveness. In this work, we critically evaluate DD methods through large-scale experiments using standardized datasets and evaluation protocols to assess their intrinsic effectiveness. We benchmark seven state-of-the-art (SOTA) DD methods on ImageNet-1K, ImageNet100, and ImageNette, using three widely adopted training protocols against four CS strategies. Our results show that while some DD methods fail to outperform even simple random subsets, the SOTA DD approaches are comparable to or worse than coresets on large-scale datasets and incur a substantially higher cost for construction. Beyond accuracy, we also evaluate the representativeness, diversity, and quality of condensed sets, and find that coresets consistently achieve better coverage of the original data distribution. These findings highlight the limited practical advantages of current DD methods and show that coresets remain competitive and are often a more computationally efficient alternative for data-centric learning.

cs.LG

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.34971

SVD-Attention: Efficiently Analyzing Attention Blocks in Transformers

Debopam Sanyal, Hongjie Chen, Riya Elizabeth Kuriyan, Arth Tak, Divya Kothandaraman, Akshay Mehra, Joshua Kimball, Alexey Tumanov

原文摘要

Pretrained transformers spend a large share of their inference compute inside self-attention, yet the per-head query-key interaction is structurally low-rank by construction. We ask a diagnostic question: across layers and architectures, which heads actually use the full rank budget on real data, and which do not? We propose SVD-Attention, a post-hoc method that decomposes the per-head interaction matrix $M_h$ and scores every singular mode by both its spectral magnitude and its activation variance across tokens on a small calibration set. The score keeps the modes the head can amplify and that real data actually activates, and discards the rest; the resulting factorization is exact for the retained modes and needs no finetuning. Across seven pretrained models and five task types, the diagnostic finds a task-dependent pattern: redundancy sits in the early to middle layers for sentiment and image classification and for generation, and in the late layers for dense prediction and extractive question answering. It is also predictive: truncating the layers it flags costs at most $1.75$ F1, while truncating the same number of layers elsewhere in the same model costs up to $50$ F1. SVD-Attention is an analysis tool rather than a compression method: acting on it reduces inference FLOPs by $0.8$-$1.5\%$ on its own and by $8$-$11\%$ when composed with weight pruning, with $0$-$4\%$ change in task quality.

cs.CL · cs.AI · cs.LG

查看论文
arXiv ↗PDF ↗