新方法 Design-CP 利用上下文并行加速蛋白质纳米颗粒设计

一篇题为《Design-CP: Context Parallelism for Design of Protein Nanoparticles》的论文在 arXiv 上发布。该研究由 Lorenzo Tarricone 等三人合作完成，提出了一种名为 Design-CP 的上下文并行方法，旨在优化蛋白质纳米颗粒的设计流程。蛋白质纳米颗粒在疫苗开发、药物递送及生物材料领域具有广阔应用前景，但其计算设计过程往往计算密集且耗时。Design-CP 通过并行处理上下文信息，显著提升了设计效率，同时保持了对纳米颗粒结构特性的精准控制。作者在论文中展示了该方法在多种设计任务上的表现，验证了其在降低计算开销、加速迭代优化方面的潜力。目前该论文提供 PDF 及 HTML 版本，相关代码与数据链接已同步公开，供研究人员复现与扩展。这项成果有望推动蛋白质纳米颗粒的快速理性设计，为生物工程与纳米医学领域提供新的计算工具。 #蛋白质纳米颗粒 #DesignCP #上下文并行 #arXiv #计算生物学 #生物工程 #纳米医学 #加速设计 #并行计算

统计有意义的几何与规范对称性破缺理论提出，为智能涌现提供几何框架

近日，一篇题为《Statistically Meaningful Geometry and Gauge Symmetry Breaking: A Geometric Foundation for Scientific Discovery and Intelligence Emergence》的论文在arXiv预印本平台正式提交。该论文由Bing Cheng等三位研究人员共同撰写，提出将统计有意义的几何（Statistically Meaningful Geometry）与规范对称性破缺（Gauge Symmetry Breaking）相融合，构建一个统一的几何基础，旨在解释科学发现过程以及智能涌现现象。研究认为，这种几何框架可能为理解复杂系统中的模式形成、知识发现以及人工智能的涌现特性提供全新的理论视角。目前论文已公开，正等待学术界进一步评审与探讨。 #arXiv #几何理论 #规范对称性 #智能涌现 #科学发现 #人工智能 #理论物理 #复杂系统

有界峰度下确切最坏情况尾概率

据arXiv平台显示，一篇题为《The Exact Worst-Case Tail Probability under Bounded Kurtosis》的数学与统计学论文近日提交并更新。该论文由Xiaoyu Li与其他三位作者共同完成，深入研究了在峰度（衡量分布尾部厚度的重要指标）有界的情况下，尾概率可能达到的最坏情况精确值。通过严谨的理论推导，论文给出了显式的数学表达式，揭示了在给定峰度约束下尾概率的上界。这一成果对于概率论、风险管理、金融保险等领域具有重要的理论指导意义，有助于更好地理解和评估极端事件的发生概率。目前论文已发布第二版，并提供PDF和HTML版本供学界查阅。 #数学 #统计学 #概率论 #尾概率 #峰度 #风险管理 #学术研究 #arXiv

稳定化高阶影响函数理论论文在arXiv发布

学者Na Liu与合作者近日在arXiv预印本平台提交题为《稳定化的高阶影响函数：一类双线性形式的统计理论》的论文。该研究聚焦于影响函数（influence functions）的推广，针对双线性形式提出了一种稳定化的高阶影响函数，并系统建立了其统计理论。高阶影响函数在统计推断与模型诊断中具有关键作用，但传统方法往往受限于不稳定性。该工作从理论层面给出了一类特殊形式的收敛性与稳健性分析，有望为机器学习、计量经济学等领域的模型评估提供更严谨的数学工具。论文于2026年7月首次提交，后经一次修订，目前提供HTML、PDF及TeX源码下载。 #统计学 #影响函数 #高阶影响函数 #双线性形式 #arXiv #论文 #机器学习 #统计理论 #稳健推断

破译大语言模型解码陷阱

近日，一篇题为《Breaking the Likelihood Trap: Variance-Calibrated Modulation for Large Language Model Decoding》的论文在arXiv预印本平台发布。该研究由Yuanhao Ding等五位学者完成，针对大语言模型（LLM）解码过程中普遍存在的“似然陷阱”问题，提出了一种名为方差校准调制的新方法。该方法通过动态调整模型输出概率分布的方差，旨在避免解码陷入过度确定或重复生成的低质量状态，从而提升文本生成的多样性与准确性。论文于2026年6月首次提交，7月更新第二版，现可免费获取全文。这项研究对优化LLM的生成策略具有潜在价值，属于计算机科学与统计学的交叉探索。 #论文 #大语言模型 #解码 #LLM #arXiv #AI #方差校准 #似然陷阱

分布式数据中心能量与AIGC调度研究提出扩散辅助奖励塑形新方法

一篇提交至arXiv的学术论文提出了联合能量管理与协调AIGC工作负载调度的方法。该研究由Yang Fu等六位作者完成，针对分布式数据中心能源消耗高昂、AIGC任务调度复杂等挑战，引入基于扩散模型辅助的奖励塑形策略。该方法通过扩散过程对奖励函数进行重塑，使智能体在动态环境中更有效地平衡能源成本和计算性能。研究团队还公开了相关代码与数据，以促进学术交流和实践应用。该工作为绿色数据中心建设和高效AI计算提供了新的技术路径。 #arXiv #论文 #能量管理 #AIGC #数据中心 #扩散模型 #奖励塑形 #人工智能 #能源效率

Conformal Prediction Sets for Instance Segmentation

Focus to learn more arXiv-issued DOI via DataCite Submission history From: Kerri Lu [ view email ] [v1] Tue, 10 Feb 2026 18:15:06 UTC (2,380 KB) [v2] Tue, 7 Jul 2026 10:45:01 UTC (2,766 KB) Full-text links: Access Paper: View a PDF of the paper titled Conformal Prediction Sets for Instance Segmentation, by Kerri Lu and 3 other authors View PDF TeX Source view license Current browse context: < prev | next > new | recent | 2026-02 Change to browse by: cs stat 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? ) L

itmaps 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? ) 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 .

《Factorizable joint shift revisited》论文发表，重新审视联合漂移可分解性

Dirk Tasche 在《Foundations of Data Science》期刊上发表了题为《Factorizable joint shift revisited》的论文，该研究聚焦于机器学习中的联合漂移（joint shift）问题，重新探讨其可分解性性质。论文于2026年1月提交至arXiv预印本平台，随后经过多个版本更新，最终以早期访问形式被期刊接收。联合漂移是分布偏移的一种重要形式，在迁移学习、领域自适应等领域具有关键影响。该成果为理解联合漂移的分解条件提供了新的理论视角，有助于推动相关算法的设计与优化。 #机器学习 #分布偏移 #迁移学习 #学术论文 #arXiv #数据科学 #人工智能

Geometric Stability: The Missing Axis of Representations 论文在 arXiv 发布

arXiv 预印本平台近日更新了一篇题为《Geometric Stability: The Missing Axis of Representations》的研究论文。该论文由研究者 Prashant Raju 撰写，自 2026 年 1 月首次提交以来，经过多次修订，至 7 月已更新至第五版。论文聚焦表示学习领域，提出几何稳定性是当前模型表示中一个被忽视的关键维度，对提升深度学习模型在几何变换下的鲁棒性与泛化能力具有潜在价值。该工作已引发学术社区关注，并提供新的理论视角。 #arXiv #论文 #表示学习 #几何稳定性 #机器学习 #深度学习 #AI #学术研究

Medix: Out-of-Distribution Detection from Unlabeled Wild Data via Robust Gradient Statistics

Focus to learn more arXiv-issued DOI via DataCite Submission history From: Momin Abbas [ view email ] [v1] Tue, 7 Oct 2025 22:43:57 UTC (931 KB) [v2] Mon, 6 Jul 2026 19:59:23 UTC (2,575 KB) Full-text links: Access Paper: View a PDF of the paper titled Medix: Out-of-Distribution Detection from Unlabeled Wild Data via Robust Gradient Statistics, by Momin Abbas and Ali Falahati and Hossein Goli and Mohammad Mohammadi Amiri View PDF HTML (experimental) TeX Source view license Current browse context: < prev | next > new | recent | 2025-10 Change to browse by: cs math stat 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 Tool

s 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 .

Variational Learning of Disentangled Representations

Focus to learn more arXiv-issued DOI via DataCite Submission history From: Ozgur Beker [ view email ] [v1] Fri, 20 Jun 2025 17:36:12 UTC (46,569 KB) [v2] Fri, 12 Dec 2025 20:31:20 UTC (13,444 KB) [v3] Mon, 6 Jul 2026 22:43:33 UTC (14,115 KB) Full-text links: Access Paper: View a PDF of the paper titled Variational Learning of Disentangled Representations, by Yuli Slavutsky and 2 other authors View PDF HTML (experimental) TeX Source view license Current browse context: < prev | next > new | recent | 2025-06 Change to browse by: cs stat 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 Expl

orer? ) 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 .

选择性状态空间模型非连续门控的规则性与稳定性研究获新进展

一篇由Nikola Zubić等人提交的论文《Regularity and Stability Properties of Selective SSMs with Discontinuous Gating》已被机器学习领域权威期刊Transactions on Machine Learning Research（TMLR）接收。该论文针对具有非连续门控机制的选择性状态空间模型（SSM）展开理论分析，深入探讨了其规则性和稳定性属性。研究指出，这类模型在处理序列数据时展现出独特的动态行为，通过严格的数学推导，作者证明了在特定条件下模型能够保持稳定并表现出良好的规则性。该工作为深度学习中的高效序列建模提供了重要的理论基础，对后续优化与应用具有参考价值。论文自2025年起在arXiv上发布，历经多次修订，最终版本于2026年7月上线。 #机器学习 #状态空间模型 #TMLR #学术论文 #深度学习 #门控机制 #稳定性分析 #规则性 #AI研究

Benign Overfitting with Quantum Kernels

Focus to learn more arXiv-issued DOI via DataCite Journal reference: UAI 2026 Submission history From: Hachem Kadri [ view email ] [v1] Fri, 21 Mar 2025 10:30:42 UTC (991 KB) [v2] Tue, 7 Jul 2026 08:55:18 UTC (1,232 KB) Full-text links: Access Paper: View a PDF of the paper titled Benign Overfitting with Quantum Kernels, by Joachim Tomasi and 2 other authors View PDF HTML (experimental) TeX Source view license Current browse context: quant-ph < prev | next > new | recent | 2025-03 Change to browse by: cs stat References & Citations INSPIRE HEP 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 To

ggle 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? ) 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 .

高维逻辑回归的普适性研究及新型CGMT在数据增强中的应用

一篇发表于2025年学习理论会议（COLT）的论文深入探讨了高维逻辑回归的普适性，并提出了一种在数据依赖条件下适用的新型凸高斯极小化定理（CGMT）。该研究由Matthew Esmaili Mallory等人完成，理论框架严格证明了高维逻辑回归的统计行为在一定条件下具有普适性，即其预测性能与噪声分布的具体形式无关。同时，研究者拓展了经典CGMT工具，使其适用于协变量存在相关性的更实际场景，并从理论上验证了数据增强技术在提升模型泛化能力方面的有效性。这一工作为高维统计学习提供了新的分析工具，对机器学习理论及实际应用具有重要意义。 #高维逻辑回归 #普适性 #CGMT #数据增强 #学习理论 #COLT2025 #机器学习 #统计学习 #理论计算机科学

二元线性分类中良性过拟合的普遍性获理论证实

近期，针对高维线性分类模型“过拟合但泛化良好”的现象，arXiv上一项新研究从理论层面揭示了其普遍性。论文题为“二元线性分类中良性过拟合的普遍性”，由Ichiro Hashimoto等学者完成。研究证明，在高维设定下，线性分类器在训练误差为零时依然能对未知数据保持优异表现，这种良性过拟合并非依赖特定模型或数据分布的特例，而是线性分类方法的本质特征。研究通过严格的数学框架，统一并拓展了此前针对特定方法的结论，对理解现代机器学习中记忆与泛化的关系提供了新视角。该论文自2025年初首次提交以来已多次更新，目前仍为预印本状态。 #机器学习 #良性过拟合 #线性分类 #高维统计 #理论证明 #人工智能 #arXiv

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

研究提出部分训练三层神经网络的函数空间平均场理论

据arXiv预印本及《Journal of Machine Learning Research》2026年刊发，Zhengdao Chen等人发表论文，提出一种函数空间平均场理论，专门用于分析部分训练的三层神经网络。该理论从函数空间视角出发，通过平均场近似研究网络部分参数固定、部分训练时的动态行为，为理解浅层神经网络的训练机理与泛化特性提供了新的数学框架。相关工作已正式发表于JMLR第27卷第52期，标志着该领域理论探索的重要进展。 #机器学习 #神经网络 #平均场理论 #JMLR #深度学习 #AI #理论 #三层网络 #函数空间 #论文