Collate 协作神经网络学习

近日，一篇题为《Collate: Collaborative Neural Network Learning for Latency-Critical Edge Systems》的论文被收录于2022年IEEE第40届国际计算机设计会议（ICCD）。该研究由Shuo Huai等人提出，针对边缘系统中对延迟高度敏感的应用场景，设计了一种协作神经网络学习框架。通过让多个边缘设备协同训练与推理，Collate 在保证模型精度的同时，显著降低了端到端响应延迟，为实时性要求严苛的物联网、自动驾驶等场景提供了可行方案。论文已上线arXiv预印本平台，并公开了相关代码与数据资源。 #边缘计算 #神经网络 #协作学习 #延迟优化 #ICCD #IEEE #AI #物联网 #工业计算

研究提出辅助游戏的可证明最优学习算法

一篇题为“Provably Optimal Learning Algorithms for Assistance Games”的学术论文近日在arXiv预印本平台发布。该论文由Nivasini Ananthakrishnan等五位研究者共同完成，聚焦于人工智能对齐领域中的辅助游戏（Assistance Games）框架。研究团队提出了可证明最优的学习算法，从理论上保证了算法在辅助游戏场景中的性能最优性，为AI系统如何更有效地协助人类提供了坚实的理论基础。该成果属于计算机科学范畴，对推动人机协作与AI安全研究具有重要参考价值。 #论文 #arXiv #辅助游戏 #学习算法 #AI #计算机科学 #人机协作 #人工智能对齐

KronQ: LLM Quantization via Kronecker

来自arXiv的最新论文提出了一种名为KronQ的大语言模型量化方法。该方法利用Kronecker分解的Hessian矩阵来优化量化过程，在保持模型性能的同时显著降低计算资源需求。研究团队通过实验验证，该技术能有效压缩LLM的权重和激活值，减少内存占用和推理延迟，为在边缘设备上部署大型语言模型提供了新思路。论文作者包括Donghyun Lee等人，于2026年7月提交，相关代码和数据已在GitHub等平台公开。 #LLM #量化 #Kronecker #Hessian #深度学习 #模型压缩 #AI #论文 #arXiv

线性注意力架构

一项关于线性注意力架构的最新研究以预印本形式发表于arXiv。该论文由Tommaso Cerruti等五位作者共同完成，系统梳理了线性注意力机制的核心原理，深入分析了其在计算效率与模型性能之间的权衡关系。论文重点提出了一种新颖的跨层路由方法，旨在优化注意力分布在不同网络层间的传递，从而降低传统Transformer模型在处理长序列时的二次复杂度。这项研究为设计更高效、可扩展的深度学习架构提供了理论依据和实验参考，有望推动自然语言处理、计算机视觉等领域的轻量化模型发展。 #人工智能 #机器学习 #注意力机制 #线性注意力 #Transformer #深度学习 #论文 #arXiv

评估基础模型在极端环境事件中的泛化能力

一篇发表在arXiv上的研究论文评估了基础模型在极端环境事件中的泛化能力，以加利福尼亚州野火导致的PM2.5污染为案例。研究由Yongcan Huang等人完成，旨在检验当前先进的基础模型能否有效预测并泛化至罕见且剧烈的环境灾害场景。加州野火频发，其产生的PM2.5浓度波动剧烈，对空气质量模型提出严峻挑战。论文通过对比多类基础模型在历史野火数据上的表现，分析了模型在不同时空条件下的预测鲁棒性。研究发现，现有基础模型在正常天气条件下表现良好，但在极端野火事件中泛化能力显著下降，亟需针对极端分布进行专门优化。该工作为将人工智能应用于环境应急响应提供了关键参考，也揭示了基础模型在气候变化适应中的局限性。 #arXiv #基础模型 #环境事件 #野火 #PM2.5 #泛化能力 #机器学习 #加州 #气候变化 #论文

path_boost：基于路径梯度提升的可解释图预测Python包

近日，研究人员在arXiv上发布了一篇论文，介绍了一款名为path_boost的Python包，该包专注于基于路径的梯度提升方法，实现可解释的图级别预测。该工具由Claudio Meggio等人开发，旨在为图结构数据提供一种高效且可解释的预测方案。路径梯度提升方法通过提取图中的路径特征，结合梯度提升框架，在保持预测性能的同时增强了模型的可解释性。该包适用于化学、社交网络、生物信息学等多个领域的图分析任务，为研究人员和开发者提供了便捷的解决方案。论文已提交至arXiv预印本平台，并提供了完整的PDF和代码链接。 #Python #机器学习 #图预测 #梯度提升 #可解释AI #arXiv #开源 #学术论文

Telegram必备的搜索引擎，极搜JISOU帮你精准找到，想要的群组、频道、视频、音乐

👉 t.me/jisou2?start=a_8247614025

高效安全对齐

近日，一篇题为《Efficient Safety Alignment of Language Models via Latent Personality Traits》的论文在arXiv上提交。该研究由Mohamed Amine Merzouk等五位学者共同完成，提出通过挖掘语言模型中的潜在人格特质来实现高效的安全对齐。传统方法依赖大量人工标注与计算资源，而该工作借鉴人格心理学中的潜在特质模型，以更低成本引导模型行为符合安全规范。这一方法有望提升AI系统的可控性与安全性，为大模型的安全部署提供新思路。论文于2026年7月8日提交，目前可在arXiv上获取全文。 #AI安全 #语言模型 #人格特质 #arXiv #论文 #机器学习 #安全对齐 #大模型

多模态遗忘学习综述：覆盖视觉、语言、视频与音频的方法、数据集与基准

近日，一篇由Nobin Sarwar等人撰写的综述论文在arXiv上发布，系统梳理了多模态遗忘学习（Multimodal Unlearning）领域的研究进展。该论文广泛覆盖视觉、语言、视频和音频等多种模态，全面总结了当前主流的遗忘方法、公开数据集以及评估基准。随着大模型在多模态处理中的广泛应用，数据隐私、模型鲁棒性和合规性成为关键挑战，遗忘学习技术旨在让模型有选择地“忘记”特定数据，从而保障安全与伦理。这篇综述为研究者提供了系统的技术全景，有助于推动该领域标准化与后续发展。 #多模态遗忘学习 #AI安全 #机器学习 #arXiv论文 #综述 #数据隐私 #模型鲁棒性

Distributed Sketching on Data Partitions for OLS Regression

Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Chao Lan [ view email ] [v1] Wed, 8 Jul 2026 19:50:45 UTC (363 KB) Full-text links: Access Paper: View a PDF of the paper titled Distributed Sketching on Data Partitions for OLS Regression, by Luyuan Yang and 3 other authors View PDF HTML (experimental) TeX Source view license Current browse context: < prev | next > new | recent | 2026-07 Change to browse by: cs References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer ( What is the Explorer? ) Connected Papers Toggle Connected Papers ( What is Connected Pa

pers? ) Litmaps Toggle Litmaps ( What is Litmaps? ) Toggle scite Smart Citations ( What are Smart Citations? ) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv ( What is alphaXiv? ) Links to Code Toggle CatalyzeX Code Finder for Papers ( What is CatalyzeX? ) DagsHub Toggle DagsHub ( What is DagsHub? ) GotitPub Toggle ( What is GotitPub? ) Huggingface Toggle Hugging Face ( What is Huggingface? ) ScienceCast Toggle ScienceCast ( What is ScienceCast? ) Demos Demos Replicate Toggle Replicate ( What is Replicate? ) Spaces Toggle Hugging Face Spaces ( What is Spaces? ) Spaces Toggle ( What is ? ) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower ( What are Influence Flowers? ) Core recommender toggle CORE Recommender ( What is CORE? ) IArxiv recommender toggle IArxiv Recommender ( What is IArxiv? ) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

Optimal Learning Rate Scaling Depends on Data in Deep Scalar Linear Networks

Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Yedi Zhang [ view email ] [v1] Wed, 8 Jul 2026 19:39:11 UTC (459 KB) Full-text links: Access Paper: View a PDF of the paper titled Optimal Learning Rate Scaling Depends on Data in Deep Scalar Linear Networks, by Yedi Zhang and 3 other authors View PDF HTML (experimental) TeX Source view license Current browse context: < prev | next > new | recent | 2026-07 Change to browse by: cs References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer ( What is the Explorer? ) Connected Papers Toggle Conn

ected Papers ( What is Connected Papers? ) Litmaps Toggle Litmaps ( What is Litmaps? ) Toggle scite Smart Citations ( What are Smart Citations? ) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv ( What is alphaXiv? ) Links to Code Toggle CatalyzeX Code Finder for Papers ( What is CatalyzeX? ) DagsHub Toggle DagsHub ( What is DagsHub? ) GotitPub Toggle ( What is GotitPub? ) Huggingface Toggle Hugging Face ( What is Huggingface? ) ScienceCast Toggle ScienceCast ( What is ScienceCast? ) Demos Demos Replicate Toggle Replicate ( What is Replicate? ) Spaces Toggle Hugging Face Spaces ( What is Spaces? ) Spaces Toggle ( What is ? ) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower ( What are Influence Flowers? ) Core recommender toggle CORE Recommender ( What is CORE? ) IArxiv recommender toggle IArxiv Recommender ( What is IArxiv? ) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

Physics-Informed Machine Learning Under Small-Data Constraints: Lessons from Abrasive Waterjet Milling

Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Jörg Frochte [ view email ] [v1] Wed, 8 Jul 2026 18:54:09 UTC (41 KB) Full-text links: Access Paper: View a PDF of the paper titled Physics-Informed Machine Learning Under Small-Data Constraints: Lessons from Abrasive Waterjet Milling, by Sarah Grewe and 1 other authors View PDF HTML (experimental) TeX Source view license Current browse context: < prev | next > new | recent | 2026-07 Change to browse by: cs References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer

( What is the Explorer? ) Connected Papers Toggle Connected Papers ( What is Connected Papers? ) Litmaps Toggle Litmaps ( What is Litmaps? ) Toggle scite Smart Citations ( What are Smart Citations? ) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv ( What is alphaXiv? ) Links to Code Toggle CatalyzeX Code Finder for Papers ( What is CatalyzeX? ) DagsHub Toggle DagsHub ( What is DagsHub? ) GotitPub Toggle ( What is GotitPub? ) Huggingface Toggle Hugging Face ( What is Huggingface? ) ScienceCast Toggle ScienceCast ( What is ScienceCast? ) Demos Demos Replicate Toggle Replicate ( What is Replicate? ) Spaces Toggle Hugging Face Spaces ( What is Spaces? ) Spaces Toggle ( What is ? ) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower ( What are Influence Flowers? ) Core recommender toggle CORE Recommender ( What is CORE? ) IArxiv recommender toggle IArxiv Recommender ( What is IArxiv? ) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

NFTR: From Provable Mode-Averaging to Geodesic Subgoal Selection in Offline Goal

Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Erdemt Bao [ view email ] [v1] Wed, 8 Jul 2026 18:33:48 UTC (1,043 KB) Full-text links: Access Paper: View a PDF of the paper titled NFTR: From Provable Mode-Averaging to Geodesic Subgoal Selection in Offline Goal-Conditioned RL, by Erdemt Bao and 2 other authors View PDF HTML (experimental) TeX Source view license Current browse context: < prev | next > new | recent | 2026-07 Change to browse by: cs References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer ( What is the Explorer? ) Con

nected Papers Toggle Connected Papers ( What is Connected Papers? ) Litmaps Toggle Litmaps ( What is Litmaps? ) Toggle scite Smart Citations ( What are Smart Citations? ) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv ( What is alphaXiv? ) Links to Code Toggle CatalyzeX Code Finder for Papers ( What is CatalyzeX? ) DagsHub Toggle DagsHub ( What is DagsHub? ) GotitPub Toggle ( What is GotitPub? ) Huggingface Toggle Hugging Face ( What is Huggingface? ) ScienceCast Toggle ScienceCast ( What is ScienceCast? ) Demos Demos Replicate Toggle Replicate ( What is Replicate? ) Spaces Toggle Hugging Face Spaces ( What is Spaces? ) Spaces Toggle ( What is ? ) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower ( What are Influence Flowers? ) Core recommender toggle CORE Recommender ( What is CORE? ) IArxiv recommender toggle IArxiv Recommender ( What is IArxiv? ) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

When Does Continual Learning Require Learning

Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Yutong Bai [ view email ] [v1] Wed, 8 Jul 2026 18:27:40 UTC (6,026 KB) Full-text links: Access Paper: View a PDF of the paper titled When Does Continual Learning Require Learning, by Anne Harrington and 9 other authors View PDF HTML (experimental) TeX Source view license Current browse context: < prev | next > new | recent | 2026-07 Change to browse by: cs References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer ( What is the Explorer? ) Connected Papers Toggle Connected Papers ( What is Connected Papers? ) Litmaps Togg

le Litmaps ( What is Litmaps? ) Toggle scite Smart Citations ( What are Smart Citations? ) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv ( What is alphaXiv? ) Links to Code Toggle CatalyzeX Code Finder for Papers ( What is CatalyzeX? ) DagsHub Toggle DagsHub ( What is DagsHub? ) GotitPub Toggle ( What is GotitPub? ) Huggingface Toggle Hugging Face ( What is Huggingface? ) ScienceCast Toggle ScienceCast ( What is ScienceCast? ) Demos Demos Replicate Toggle Replicate ( What is Replicate? ) Spaces Toggle Hugging Face Spaces ( What is Spaces? ) Spaces Toggle ( What is ? ) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower ( What are Influence Flowers? ) Core recommender toggle CORE Recommender ( What is CORE? ) IArxiv recommender toggle IArxiv Recommender ( What is IArxiv? ) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

Explaining Near-Zero Hessian Eigenvalues Through Approximate Symmetries in Neural Networks

Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Marcel Kühn [ view email ] [v1] Wed, 8 Jul 2026 18:27:16 UTC (1,239 KB) Full-text links: Access Paper: View a PDF of the paper titled Explaining Near-Zero Hessian Eigenvalues Through Approximate Symmetries in Neural Networks, by Marcel K\"uhn and 1 other authors View PDF HTML (experimental) TeX Source view license Current browse context: < prev | next > new | recent | 2026-07 Change to browse by: cond-mat -nn cs References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (

What is the Explorer? ) Connected Papers Toggle Connected Papers ( What is Connected Papers? ) Litmaps Toggle Litmaps ( What is Litmaps? ) Toggle scite Smart Citations ( What are Smart Citations? ) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv ( What is alphaXiv? ) Links to Code Toggle CatalyzeX Code Finder for Papers ( What is CatalyzeX? ) DagsHub Toggle DagsHub ( What is DagsHub? ) GotitPub Toggle ( What is GotitPub? ) Huggingface Toggle Hugging Face ( What is Huggingface? ) ScienceCast Toggle ScienceCast ( What is ScienceCast? ) Demos Demos Replicate Toggle Replicate ( What is Replicate? ) Spaces Toggle Hugging Face Spaces ( What is Spaces? ) Spaces Toggle ( What is ? ) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower ( What are Influence Flowers? ) Core recommender toggle CORE Recommender ( What is CORE? ) IArxiv recommender toggle IArxiv Recommender ( What is IArxiv? ) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .