技术博客与 newsletter feed

DEV Community · Top · Week · 科技

27 个榜单12分钟前更新默认榜单
共 30 条热搜查看完整榜单
共 30 条热搜查看完整榜单
共 30 条热搜查看完整榜单
共 18 条热搜查看完整榜单
共 413 条热搜查看完整榜单
共 9 条热搜查看完整榜单
共 27 条热搜查看完整榜单
共 143 条热搜查看完整榜单
  • 01
    Why Don't More Developers 'Use the Platform'?
    A thoughtful look at why we often rebuild what browsers already offer, from someone usually on the platform's side. It kicked off an extensive Hacker News discussion , with many arguing native controls and APIs are too limited or awkward to depend on.Nolan Lawson
  • 02
    Preact 11 Released
    Six years after the road to Preact 11 began, the latest version of the popular "alternative to React with the same modern API" is here, with improved hydration, automatic ref forwarding, and an ESM-only package. The upgrade guide covers the changes.The Preact Team
  • 03
    Nuxt 4.6: A Big Step Towards a Server-Agnostic Nuxt
    A meaty release for the full-stack Vue framework with a server layer that no longer ties you to Nitro (pure Vite server builds are now an experimental option), built-in sessions, Vue Vapor support, and a new CLI ( Nuxt CLI 4.0 ).Daniel Roe
  • 04
    Getting Started with Claude Code Mods
    Claude Code has just added support for Mods , a way to extend the agentic coding harness with JavaScript so you can watch events, change behavior, and draw custom UI. Addy builds a mod here in ~80 lines that shows a live 'forecast' of your context window.Addy Osmani
  • 05
    Friendship Ended with Deno, Now Node is My Best Friend
    David felt good about Deno in 2021, but has now returned to Node and is surprised by how much has improved. He ported his site generator back with barely any changes, and doesn't hold back on where he thinks Deno went wrong.David Bushell
  • 06
    I Tested 11 HTTP Resilience Libraries
    How do fetch wrappers like Ky , ofetch , and Wretch behave if you abort mid-retry? The author of ffetch tested eleven options and found five still attempt retries after cancellation.Gabor Koos
  • 07
    TanStack Charts 1.0: A Composable, Typed Charting Library
    A framework-neutral way to declaratively specify visualizations in typed, tree-shakable JavaScript which are then rendered with SVG or Canvas. The extensive sample catalog (with code) shows off line, bar, pie and area charts, plus heatmaps, box plots, histograms, etc.Tanner Linsley
  • 08
    SvelteKit 3: 'The Same Framework with a Little More Polish'
    Svelte's official app framework gets a cleanup/polish release. Few new features, and remote functions remain experimental, but existing users have some work to do to migrate (though sv migrate can handle much of it ).The Svelte Team
  • 09
    Effect 4.0: A Smaller Core, Rebuilt Runtime
    A ground-up rewrite of the popular library for building reliable apps that handles tricky stuff like resource management, concurrency, retries, typed errors, and more. v4.0 boasts 5x smaller minimal bundles and 86% less memory used per concurrent task.Sebastian Lorenz
  • 10
    Lightpanda 1.0: A Headless Browser for Automation
    A headless browser built from scratch in Zig. With no graphical rendering, it uses a fraction of Chrome's CPU and memory (~35MB memory in a run I just did!). Drive it with Puppeteer/Playwright or, the fun bit, script it with plain JavaScript .Francis Bouvier
共 10 条热搜查看完整榜单
  • 01
    高分辨率植被覆盖度遥感产品及算法研究进展
    植被覆盖度FVC（fractional vegetation cover）是表征植被冠层结构与地表植被覆盖状况的关键生物物理参数之一。近年来，随着遥感数据时空分辨率的提升，FVC遥感估算方法取得显著进展，基于相关算法生产的FVC产品及其数据集已广泛应用于气候变化与生态环境研究。然而，现有主流FVC遥感产品多集中于中低空间分辨率，难以满足局地尺度植被精细化监测的需求。尤其在空间异质性较高的区域，易产生混合像元效应，从而增加FVC反演结果的不确定性。因此，发展高空间分辨率的FVC遥感估算方法与产品具有重要的科学意义。本文首先系统梳理了当前公开发布的典型高分辨率（≤30 m）FVC遥感产品的时空尺度特征及优缺点，对比总结了经验回归模型、光谱解混模型、物理机理模型以及机器/深度学习模型等4类典型估算方法；然后围绕产品真实性检验，分析了常见检验方法及评价指标，讨论了地面真值数据获取困难、尺度不匹配以及空间异质性影响等关键难点；最后从数据源扩展、算法提升以及一致性检验体系等方面，对未来高分辨率FVC产品发展方向进行展望，以期为高分辨率FVC估算方法优化与产品研发提供理论参考。杜晓铮, 赵祥, 贾坤, 赵嘉诚
  • 02
    作物生长模型与遥感数据同化研究新进展
    数据同化DA（data assimilation）通过融合作物生长模型CGMs（crop growth models）的过程机理与遥感RS（remote sensing）观测信息，为大尺度、连续的作物状况监测与产量预测提供了关键技术支撑。本文综述了近年来该领域的研究进展，重点从同化变量拓展和算法体系演进两个方面展开论述。在同化变量方面，分析了日光诱导叶绿素荧光SIF（solar-induced chlorophyll fluorescence）、植被光学深度VOD（vegetation optical depth）、合成孔径雷达SAR（synthetic aperture radar）数据在表征作物生理状态、冠层结构和水分胁迫方面的独特优势，并对目前存在的问题及进一步发展进行了讨论。在算法方面，总结了混合滤波方法在高维体系中的适用性，以及以机器学习ML（machine learning）为代表的代理模型在提高同化计算效率方面发挥的重要作用，同时指出ML在部分研究中也被用于改进协方差矩阵估计。此外，本文结合“可同化性”（Assimilability）框架，从模型结构、参数可识别性和观测算杜燕飞, 蒙继华, 林圳鑫, 何荣鹏, 高心雨, 那浩宇
  • 03
    嫦娥六号微型激光角反射器阵列设计与分析
    微型激光角反射器阵列INRRI（INstrument for landing-Roving laser Retroreflector Investigations）是嫦娥六号任务搭载的国际载荷之一，通过接收环月轨道器搭载激光设备发射的激光，利用激光入射到反射面后平行反射的特性实现高精度测距，并经过重复观测与解算成为月球背面首个绝对控制点，为月球大地测量、遥感制图与定位、环月轨道器高精度定轨和导航等提供基础保障。为探究微型激光角反射器阵列在月球环境下的工作特性与工程应用能力，本文以INRRI作为研究对象，介绍其机械结构与光学设计，并对其关键性能进行了分析。在设计层面，采用了球形穹面设计以及镀膜技术，使INRRI能够在更广阔的视场范围内有效接收轨道器激光；在性能分析层面，本文建立了一个有效反射面积模型，分析在不同入射角度下的INRRI的回波信号强度，并结合速差效应与远场衍射理论，综合评估INRRI对环月轨道器搭载激光设备的可探测性。结果表明，INRRI具备良好的可观测性，验证了其经环月轨道器LRO（Lunar Reconnaissance Orbiter）搭载的激光高度计LOLA（Luna周天豪, 王晔昕, Dell’agnello Simone, 邸凯昌, 陈少华, Lorenzo Salvatori, 郭广妍, Mattia Tibuzzi, 陈强龙, Raffaele Rodriquez, Roberto Campagnola, Rudi Lauretani, 周雅诵, Blanca Villalba, 陈天昊
  • 04
    融合空谱注意力机制的SSA-UNet沙尘检测方法研究——以FY-4A卫星数据为例
    高精度沙尘检测对沙尘传输研究、气候评估及大气环境研究具有重要意义。本研究基于国产FY-4A卫星搭载的先进的静止轨道辐射成像仪AGRI（Advanced Geostationary Radiation Imager）观测数据，结合ERA-5再分析资料等数据，提出了一种融合空谱注意力机制的改进U-Net沙尘检测模型SSA-UNet（Spatial-Spectral Attention U-Net）。在模型输入方面，考虑了沙尘的光谱特性与空间分布特征，构建了包含多波段反射率、亮温观测、沙尘光谱指数以及气象参数的多维特征输入。该模型通过在U-Net架构中引入通道和空间双注意力模块，可显著提升模型对沙尘空间和光谱特征的提取能力；采用残差跳跃连接结构，可有效缓解深层网络的梯度退化问题。在验证集上的验证结果表明，SSA-UNet的整体准确率达到99.46%，平均交并比为93.57%，F1-score为87.17%，验证精度显著优于传统的物理指数方法和仅依赖空间信息的机器学习方法。在独立沙尘事件检测测试中（数据未参与训练），SSA-UNet的沙尘识别结果与目视解译结果的总体准确率达到98.77%，进一冯自贤, 佘璐, 杨军明, 闫琛, 赵红红
  • 05
    风云三号降水星微波成像仪地理定位与偏差校正
    风云三号G星（FY-3G）是中国首颗主动降水测量卫星，其上搭载的微波成像仪—降水型MWRI-RM（micro-wave radiation imager-rainfall mission）全新升级换代，在国内首次实现了17频点26通道一体化探测。微波遥感数据地理位置偏差会直接影响后续多源产品的融合精度，引入不易剥离的误差。为了提高全新MWRI-RM的定位精度，本研究针对仪器的圆锥扫描设计体制，推导了精密矢量观测几何模型，分析了圆锥型扫描微波图像沿轨、跨轨和扫描角在轨定位偏差特征和转换关系，并提出了两种定位偏差的表征方法。基于海岸线刃边匹配算法的定位检验结果，进一步推导了从物方至像方的等效失配角偏差订正模型。在轨测试数据分析表明，精密矢量观测几何模型相比传统视角模型能够显著减小MWRI-RM在轨初期的定位偏差。进一步通过偏差订正模型调整安装参数后，在轨业务稳定阶段MWRI-RM的沿轨定位精度为（-0.21±4.71） km，跨轨定位精度为（0.15±1.42） km，达到全球5 km定位指标要求，全视场整体达到1像元，满足了实时业务和产品需求。王嘉琛, 杨磊, 商建, 刘成保, 赵兴辉, 武胜利, 姚蓬娟
  • 06
    静止轨道微波探测卫星——风云四号微波星垂直探测仪GeoMWS仿真资料的海洋云检测
    中国研发的静止轨道微波探测卫星——风云四号微波星垂直探测仪GeoMWS（Fengyun-4 Geostationary Microwave Sounder），可以提供高时间分辨率的微波观测资料，对天气系统的监测和数值天气预报具有重要作用。云检测是数值预报系统卫星资料同化中的关键前处理环节，但是已有的微波探测仪云检测方法多是基于极轨微波探测仪器特征设计，因此开发适用于静止微波探测仪的云检测方法是该卫星资料同化应用的前提条件。鉴于此，本研究基于GeoMWS仪器的仿真资料，分析了各通道亮温随云深厚程度的变化特征，并在此基础上提出了一种适用于GeoMWS的快速云检测方法——静止轨道微波探测仪云检测方法GeoCDM（Geostationary Microwave Sounder Cloud Detection Method），并利用长时间的仿真资料进行了定量评估。评估结果表明，该方法在不同时间段内均表现出较高的检测性能，云检测率（POD_cld）和报中率（HR）均超过75%，虚警率（FAR_cld）控制在20%以下，而已有方法的检出率仅约60%。本研究提出的GeoCDM，可为GeoMWS在台风监陶辉敏, 秦正坤, 韩阳, 胡菊旸, 毕研盟
  • 07
    中国城市植被物候的梯度差异规律及其格局
    植被物候是气候条件、下垫面变化及人类活动干扰等对生态系统的综合反映，亦是城市化和气候变化的重要反馈指标。伴随城市化进程的加快，城市与乡村的植被物候往往会呈现差异，然而有关中国城市植被物候的梯度差异规律和格局尚缺乏系统性认识。为此，本文基于哨兵2号（Sentinel-2）数据，采用动态阈值法提取2019年—2024年中国128个城市及其周边区域的植被生长季开始时间SOS（start of season）和生长季结束时间EOS（end of season），探究“城市—城镇—乡村”植被物候差异及其分布格局。研究结果表明：（1）中国城市植被SOS较周边城镇提前1.26 d，较周边乡村提前1.49 d；城市EOS较周边城镇推迟1.51 d，较周边乡村推迟1.25 d。（2）不同植被类型中，森林对城市化的物候响应最为显著且响应幅度最大。（3）不同气候背景下，温带气候区对城市化梯度的物候响应最强；亚热带和热带气候区的城乡物候差异整体较弱且稳定性较低。（4）不同城市规模下，SOS提前效应随城市规模增大而增强；EOS在中小规模城市中表现为推迟效应，在超大城市中则转变为提前效应。上述结果为城市植被物候变崔颖, 陈云浩, 耿昊, 李康宁, 李晓慧
  • 08
    融合趋势—周期特征的中国云南省森林扰动时空动态监测研究
    森林扰动精准检测是评估森林生态系统健康状况、解析碳循环动态及维护生物多样性的重要技术支撑。本研究构建了集成趋势分解与周期校正的森林扰动解析模型，该模型基于高密度Landsat卫星归一化植被指数NDVI（normalized difference vegetation index）时间序列数据提取趋势—周期双分量特征，并耦合时频域多维约束机制，准确校正了时序边缘效应，有效抑制了伪扰动信号，提高了森林扰动检测精度，同时进一步结合扰动前后土地利用动态变化特征，实现了森林扰动类型信息的准确识别。研究结果表明：（1）模型扰动检测总体精度达85.41%（Kappa为0.85，均方根误差RMSE（root mean square error）为3.81年），较全球森林变化监测数据集GFW（global forest watch）精度提升18.93%，且与基于Landsat的扰动与恢复趋势监测LandTrendr（Landsat-based detection of trends in disturbance and recovery）方法和连续变化检测与分类CCDC（continuous chang何林蒴, 周泽同, 王金亮, 王成, 黄祥, 程峰
  • 09
    有限样本稳定分类理论的新疆棉花遥感制图研究
    新疆是中国最大的产棉区，绘制新疆棉花种植分布图对于棉花生产管理具有重要意义。机器学习遥感制图精度依赖样本数量和质量，新疆棉花种植范围广，采样成本高，难以满足高精度动态制图需求。为解决这一问题，本文基于有限样本稳定分类理论，结合新疆高空间异质性特点，针对不同棉花产区（北疆、南疆和东疆棉区）通过增加地理空间约束进行样本集构建和抽样设计，从最小样本量和错误样本容忍度2个维度探讨有限样本条件下棉花分类的稳定性机制。结果表明：（1）在北疆、南疆和东疆棉区分别使用占各自样本集16%、24%和14%的样本，或将样本集的错误样本比例控制在25%、25%、10%以内，分类精度保持稳定（精度下降<1%）；（2）基于最小样本量进行新疆棉花制图能够获得较好的制图效果，北疆、南疆和东疆的总体精度分别为90.56%、83.17%和94.57%，县域尺度棉花估算面积与已有棉花分布图强相关，决定系数R 2 （coefficient of determination）和均方根误差RMSE（Root Mean Square Error）分别为0.98和5384.42 hm 2 ；（3）将样本集迁移至棉花种植面积年际变化小蓝乐淘, 黄长平, 周峻如, 曾如冰, 张泽, 张立福
  • 10
    融合频域感知与对比学习的无人机影像森林全要素解译方法：FSC-Mask2Former
    单木尺度的树种分布与生境信息是森林生态系统科学管理的重要基础，无人机可见光（RGB）影像具备采集时间灵活、空间分辨率高、获取成本低等优势，为精细尺度的森林监测提供了数据支撑。高空间分辨率影像完整记录林木的精细轮廓与森林的背景生境，利用全景分割技术对其进行统一解译，能够同步获取森林全要素的提取结果。然而，高空间分辨率影像在高郁闭度森林场景下的全景分割主要存在两方面难点：一是不同树种仅依靠的光谱信息区分度有限；二是传统方法多采用语义与实例分割任务分离的架构。导致前景与背景上下文利用不足，容易引发像素归属冲突。针对上述问题，本研究提出了一种端到端全景分割模型FSC-Mask2Former（Frequency and Supervised Contrastive Mask2Former）。该模型在统一的掩膜分类范式下进行了2项核心改进：（1）引入频域纹理感知注意力模块，通过二维离散余弦变换在频域空间捕捉高频边缘信号，强化模型对树冠微观纹理的细粒度特征提取能力；（2）设计实例感知查询对比头，利用监督对比学习策略施加判别性约束，增加相似树种间的类间特征距离。在广西南宁高峰林场等多个研究区的验证结果杨振, 姚宗琦, 张晓丽
  • 11
    结合自适应空频融合和KME分布表征的有序SAR海况等级划分方法
    海况等级划分在船舶航行安全和航线实时规划中具有重要意义。合成孔径雷达SAR（synthetic aperture radar）具有全天时、全天候成像能力，可为海况监测提供稳定数据支撑。然而，现有研究多集中于基于海浪参数的海况反演，流程复杂且在突发海况变化下响应速度受限；同时，受SAR成像相干斑噪声干扰、海况类内差异性大和等级序数特性未被充分考虑等因素影响，分类精度降低。针对上述问题，本文提出了一种基于自适应空频融合和核均值嵌入KME（kernel mean embedding）分布表征的有序SAR海况等级划分方法。首先，自适应空频融合网络增强了对图像空间域和频率域的特征表达，实现了对有效海浪信息的增强和斑点噪声的抑制。其次，KME通过将特征映射至再生核希尔伯特空间进行分布表征，以抑制类内差异性大对等级划分的影响。最后，有序分类模块按等级顺序进行阈值预测，并根据累积概率确定海况等级。此外，三重联合损失从多约束角度联合优化模型训练。在本文构建的Sentinel-1和Gaofen-3海况数据集上，所提方法相比最优方法总体精度分别提升了2.4%和2.3%，跨等级误判率降至0，实现了100%的±朱蕊, 张天文, 高贵, 吴小丹
  • 12
    艾比湖水体透明度动态变化监测与影响因素分析
    艾比湖是新疆维吾尔自治区（以下简称“新疆”）最大的咸水湖，也是西北地区重要的生态屏障，其水体透明度SDD（Secchi disk depth）是反映湖泊光学特性、生态环境状况以及响应自然与人类活动的关键指标。针对艾比湖持续萎缩的现状，以及传统SDD监测方法存在的成本高、覆盖面有限和操作风险大等问题，本研究基于2023年实测SDD数据和2010年—2024年多源卫星遥感数据，筛选出对SDD最敏感的光谱特征——绿光波段反射率与红光波段反射率的差值（R G -R R ，r=0.81），构建了适用于艾比湖的SDD遥感反演指数模型（y=7.174e 18.629 x ）。经验证，该模型的决定系数（R 2 ）为0.753，均方根误差RMSE（root mean square error）为3.7 cm，平均绝对百分比误差MAPE（mean absolute percentage error）为10.94%。利用该模型，本研究分析了近15年艾比湖SDD的时空变化特征及影响因素。结果表明：（1）艾比湖SDD整体偏低（多年均值13 cm），年际变化与水体面积呈显著正相关（r=0.59）；季节上表现为春季朱涛, 李微, 于烨铭, 王靖茹, 刘诗琦, 孙建富, 杨长乐, 孟克巴雅尔
  • 13
    联合ICESat-2与多光谱遥感反演浅水地形算法研究
    近海岸水下地形作为陆海过渡带的重要基础信息，其高效获取对于海岸带管理、岛礁工程建设等具有重要意义。ICESat-2卫星凭借光子计数激光雷达的全球覆盖能力，为浅海测深提供了新的技术途径，但基于ATL03光子点云数据的水深反演仍面临噪声密度高、光子沿轨分布非平稳及深水信号稀疏导致的提取不连续等挑战，同时对海面起伏的简化处理也限制了折射校正精度。针对上述问题，本文提出一种融合阶梯分段阈值与最大重叠离散小波变换MODWT（maximal overlap discrete wavelet transform）多尺度趋势约束的分级去噪方法。该方法首先利用高斯拟合确定海面高程范围，并结合沿轨阶梯分段策略实现水面与水下光子的初步分离；随后，采用MODWT对水面光子进行多尺度拟合，获取瞬时海面起伏信息，并将其作为折射校正的约束条件。在水下光子处理阶段，先通过直方图压缩与自适应阈值完成粗去噪，再以MODWT提取的水下地形低频趋势为先验约束进行迭代过滤，同时引入光子加密策略以增强深水稀疏信号的保留能力，最后经滑动窗口局部统计完成精去噪，从而获得连续、稳定的水下地形光子。本文以圣克鲁斯岛、阿尔达布拉环礁与塞班魏天佑, 张双成, 李军, 王涛, 王铭辉, 冯智杰, 万子涵
  • 14
    结合双原型和状态空间模型的滨海湿地跨域少样本无人机高光谱图像分类
    对滨海湿地地物进行精准识别，掌握其空间分布格局，对于推进滨海湿地的保护与修复具有重要意义。然而，由于滨海湿地地物类型多样、空间分布复杂，且实地调查与人工标注成本较高，导致高质量标注样本相对有限，进而制约了模型的分类精度。利用标记样本充足的公开数据集，结合跨域少样本方法，是缓解目标场景标注不足问题的一种有效途径，但目前的跨域少样本方法普遍缺乏先验知识的引导，且特征提取网络计算复杂度较高。为此，本文提出了一种基于双原型和状态空间模型的滨海湿地跨域少样本分类方法。首先，设计了基于状态空间模型的空间光谱特征提取器，以线性复杂度提取全局空间光谱特征；其次，结合空间分布、光谱特征与关键类别差异先验知识，通过预训练语言模型生成类别级文本原型，并与相应图像原型显式对齐，构建了语义先验驱动的文本—图像双原型框架；此外，为了更好地对齐源域和目标域分布，采用基于样本加权的域对齐方式，使网络更加关注对齐困难的样本。结果表明，在黄河三角洲滨海湿地的3个无人机高光谱图像中，在每类选用5个样本进行训练的条件下，本文方法总体精度达到了（89.12±1.74）%、（88.21±0.36）%和（78.85±0.88）%，辛紫麒, 李忠伟, 于启星, 许明明, 任广波, 王雷全, 胡亚斌, 赵旦
  • 15
    融合多重注意力机制的遥感影像小目标检测方法
    针对合成孔径雷达SAR（synthetic aperture radar）遥感小目标图像检测中目标尺度差异大、特征提取和融合困难等问题，本文提出一种融合多重注意力机制的改进YOLO11n（you only look once 11n）方法。首先，在主干网络的C3K2模块中，嵌入多尺度特征膨胀残差MFDR（multi-scale feature dilated residual）模块替换原有Bottleneck模块，利用4种不同空洞率的空洞卷积增强多尺度特征提取能力，并结合上下文聚合注意力机制模块CAAM（context aggregation attention module）实现不同特征的长距离交互，从而提取全局上下文信息。其次，设计多重注意力融合增强模块MAFEM（multi-attention fusion enhancement module），结合通道注意力与卷积增强的空间注意力机制实现多粒度融合，并采用三分支卷积RTB（three-branch convolution）模块捕获多尺度上下文信息，增强小目标及远距离目标的特征表达能力。最后，在特征融合中引入改进的特征金字塔拼接肖振久, 赵志豪, 曲海成
  • 16
    基于CNN与Mamba状态空间模型的高光谱图像压缩感知
    随着高光谱成像技术的不断发展，高光谱图像因其高维度特性，在存储与传输方面面临严峻挑战。尤其是在星载应用场景下，对在轨高光谱数据的高效压缩提出了更为迫切的需求。近年来，结合深度学习的压缩感知方法受到广泛关注，但现有方法在压缩性能、重建质量以及计算复杂度之间仍难以实现有效平衡。因此，本研究提出了一种基于Mamba状态空间选择机制与二阶段量化编码的高光谱图像压缩感知网络（Mamba-CSNet）。该网络通过引入融合卷积神经网络CNN（Convolutional Neural Network）与Mamba的特征增强模块，利用Mamba线性复杂度的状态更新机制，在保持低计算复杂度的情况下，提升对高光谱数据跨波段相关性的建模能力；同时，设计了一种二阶段量化编码策略，进一步挖掘特征冗余，以兼顾压缩效率与重建质量。实验验证结果表明，在1%采样率条件下，所提出的方法在PSNR指标上达到38.024 dB，相比当前最优方法提升了0.872 dB，同时计算量降低了11.1%。此外，在极低比特率条件（0.04—0.08 bpppb）下，得益于量化编码模块的引入，Mamba-CSNet仍能保持优异的重建质量，验陈嘉骏, 肖晶, 廖良, 王密
  • 17
    联合遥感影像配准与变化检测的轻量级网络
    影像配准和变化检测对遥感时序信息的提取与分析至关重要。当前，深度学习方法通常将二者视为独立任务，缺乏联合处理与显式配准机制。同时，现有变化检测方法缺乏差异特征空间信息与语义信息的协同交互，难以有效构建双时相影像间的实质性空谱差异。此外，二者联合模型的轻量化，对资源受限设备部署、大规模影像批处理至关重要。为此，本文提出一种联合遥感影像配准与变化检测的轻量级网络。首先，利用MobileNet V3 Large提取用于配准和变化检测的多尺度特征。其次，利用空间一致性模块实现半密集特征点匹配，并建立跨尺度的空间变换模型，使不同尺度特征图相互对齐。然后，通过时空差异协同模块增强不同尺度特征图中双时相特征的时空异质性。最后，对多尺度差异特征进行融合，生成变化检测结果。选取SVCD、SYSU-CD和SECOND数据集进行试验，将本文方法与当前主流的变化检测网络进行了对比。结果表明：本文方法能有效构建待配准影像间的空间变换关系，在定量分析和定性分析方面均优于对比方法，并在网络复杂度方面具有一定优势。龚良雄, 李星华, 程远明, 赵兴友, 龚循强, 王保国, 赵丽科, 王红根
  • 18
    面向高分辨率遥感影像变化检测的可信差异筛选与结构细化方法
    针对高分辨率遥感影像变化检测中光照、阴影、季节物候和辐射偏移引起的伪变化误检，以及建筑边界、小目标和狭长变化区域结构破碎问题，本文提出可信差异筛选与结构细化网络TGS-Net（trusted difference screening and structural refinement network）。该网络以ChangerEx双输入编码-解码框架为基础，采用IA-MixVisionTransformer（interaction-enhanced MixVisionTransformer，即交互增强型MixVisionTransformer）提取双时相多尺度特征；在融合前设计时序差分门控TDG（temporal difference gate），并经流式双重对齐融合FDAF（flow dual-alignment fusion）完成对齐融合；在融合后引入基于大选择核LSK（large selective kernel）的结构细化分支，以改善边界连续性和区域完整性。实验结果表明，在WHU-CD、LEVIR-CD和SYSU-CD数据集上，TGS-Net的F1值分别为94.58%、91.79麦超云, 李佳硕, 何海鹏, 李宏烨, 李骁, 翟懿奎, 彭志平
  • 19
    融合建筑语义信息的遥感图像道路提取方法
    从遥感图像中提取道路能够为城市规划与抢险救灾提供数据支持，在城市建设、智能交通及灾害应急等领域具有重要的应用价值。尽管基于深度学习的遥感图像道路提取方法已取得一些进展，现有模型仍面临道路错检、断裂及连通性不足等挑战。鉴于道路与建筑在空间布局上存在的强几何与拓扑关联，本文提出融合建筑语义信息的遥感图像道路提取模型，该模型能够在利用有限标注信息的情况下同时感知道路与建筑物的特征，提升预测道路结果的准确性与连通性。具体而言，首先，构建道路—建筑混合指令分割数据集，使模型可根据文本指令动态感知道路与建筑特征。其次，设计多层级指令语义注入模块，实现视觉特征与任务指令的深度融合；同时，进一步提出频率特征解耦的上下文捕捉模块，分别对图像低频特征与高频特征进行多方向建模，增强模型对遮挡与复杂背景的鲁棒性。最后，采用高斯—卷积混合解码器，协同优化全局语义一致性与局部边界精度，生成兼顾细节与连通性的分割结果。实验表明，本文所构建模型在Massachusetts与Paris数据集上均取得具有竞争力的交并比与路径长度平均相似度。定性实验验证了模型在建筑密集、阴影遮挡等场景的优势，消融实验进一步证实了建筑语义引杨志刚, 姚惠广, 田林茂, 倪维平, 吴俊政, 李强, 王琦
  • 20
    δ-M+方法的分析在矢量辐射传输模式中的应用
    大气辐射传输模拟中，强各向异性散射相函数的精确计算是关键难点。传统δ-M截断方法在处理液态云、气溶胶及冰晶等具有尖锐前向峰的粒子散射时易产生振荡，导致计算精度下降。δ-M+方法是在δ-M基础上通过以高斯加权的δ*函数替代传统狄拉克函数，使高阶勒让德系数平滑衰减，从而有效抑制系统性振荡，但δ-M+方法当前仅应用于标量辐亮度计算。本研究将δ-M+方法推广至矢量辐射传输模式计算，并在SOSVRT模式中实现了基于矩阵形式和比值缩放的2类方案。基于球形气溶胶、液态水云滴以及非球形冰晶的散射相矩阵重建应用于全矢量辐射传输仿真试验，结果表明：δ-M+方法能在保持偏振精度的同时显著提升模拟效率；与传统δ-M方法相比，在达到1%辐亮度误差条件下所需流数约其1/3。综上，在勒让德展开系数满足单调递减条件的前提下，δ-M+方法为强前向散射条件下的矢量辐射传输提供了高效且稳定的数值计算方案。周莹, 乔聪聪, 朱净淼, 周敏强, 张璐, 郭霞, 段民征
共 20 条热搜查看完整榜单
共 20 条热搜查看完整榜单
共 10 条热搜查看完整榜单
共 212 条热搜查看完整榜单
  • 01
    A Library for Learning Neural Operators
    We present NeuralOperator, an open-source Python library for operator learning. Neural operators generalize neural networks to maps between function spaces instead of finite-dimensional Euclidean spaces. They can be trained and inferenced on input and output functions given at various discretizations, satisfying a discretization convergence properties. Part of the official PyTorch Ecosystem, NeuralOperator provides all the tools for training and deploying neural operator models, as well as develJean Kossaifi, Nikola Kovachki, Zongyi Li, David Pitt, Miguel Liu-Schiaffini, Robert J. George, Boris Bonev, Kamyar Azizzadenesheli, Julius Berner, Valentin Duruisseaux, Anima Anandkumar
  • 02
    MarkDiffusion: An Open-Source Toolkit for Generative Watermarking of Latent Diffusion Models
    We introduce MarkDiffusion, an open-source Python toolkit for generative watermarking of latent diffusion models. It comprises three key components: a unified implementation framework for streamlined watermarking algorithm integration and user-friendly interfaces; a mechanism visualization suite that intuitively presents embedded and extracted watermark patterns to aid public understanding; and a comprehensive evaluation module offering standard implementations of 24 tools for assessing detectabLeyi Pan, Sheng Guan, Zheyu Fu, Luyang Si, Huan Wang, Zian Wang, Hanqian Li, Xuming Hu, Irwin King, Philip S. Yu, Aiwei Liu, Lijie Wen
  • 03
    OptunaHub: A Platform for Black-Box Optimization
    Black-box optimization (BBO) underpins advances in domains such as AutoML and Materials Informatics, yet implementations of algorithms and benchmarks remain fragmented across research communities. We introduce OptunaHub (https://hub.optuna.org/), a community-oriented, decentralized platform for distributing BBO components under a unified Optuna-compatible interface. OptunaHub enables independent publication, discovery, and reuse of optimization algorithms and benchmark problems through a lightweYoshihiko Ozaki, Shuhei Watanabe, Toshihiko Yanase
  • 04
    Unveiling the Statistical Foundations of Chain-of-Thought Prompting Methods
    Chain-of-Thought (CoT) prompting and its variants have gained significant attention as effective methods for solving multi-step reasoning tasks with pretrained large language models (LLMs). However, their theoretical underpinnings remain insufficiently explored. We analyze CoT prompting from a statistical perspective, offering insights into why “pretrained LLMs + CoT prompting” performs well. Additionally, we examine the role of the transformer architecture and the inclusion of intermediate reasXinyang Hu, Fengzhuo Zhang, Siyu Chen, Zhuoran Yang
  • 05
    Prob-GParareal: A Probabilistic Numerical Parallel-in-Time Solver for Differential Equations
    We introduce Prob-GParareal, a probabilistic extension of the GParareal algorithm designed to provide uncertainty quantification for the Parallel-in-Time (PinT) solution of (ordinary and partial) differential equations (ODEs, PDEs). The method employs Gaussian processes (GPs) to model the Parareal correction function, in line with GParareal, further enabling the propagation of numerical uncertainty across time and yielding probabilistic forecasts of the system's evolution. Furthermore, Prob-GParGuglielmo Gattiglio, Lyudmila Grigoryeva, Massimiliano Tamborrino
  • 06
    From learnable objects to learnable random objects
    We consider the relationship between learnability of a "base class" of functions on a set $X$, and learnability of a class of statistical functions derived from the base class. For example, we refine results showing that learnability of a family $h_p: p \in \Theta$ of functions implies learnability of the family of functions $h_\mu(p) = \mathbb{E}_\mu[h_p]$, where $\mathbb{E}_\mu$ is the expectation with respect to $\mu$, and $\mu$ ranges over probability distributions on $X$. We will look at boAaron Anderson, Michael Benedikt
  • 07
    A Theoretical Framework for Masked Pretraining (MPT)
    Recently, Masked Pretraining (MPT) based on reconstruction pretraining tasks has risen to a promising self-supervised learning paradigm across various domains and achieves remarkable performance in multiple downstream tasks. However, the theoretical understanding of the working mechanism behind MPT is still limited. In this paper, we introduce a new theoretical framework to analyze MPT and understand the crucial role of masking in extracting meaningful representations. We establish theoretical cQi Zhang, Runyu Zhou, Yifei Wang, Yisen Wang
  • 08
    Robustness Against Weak or Invalid Instruments: Exploring Nonlinear Treatment Models with Machine Learning
    We discuss causal inference for observational studies with possibly invalid instrumental variables. We propose a novel methodology called two-stage curvature identification (\texttt{TSCI}) by exploring the nonlinear treatment model with machine learning. The first-stage machine learning enables improving the instrumental variable's strength and adjusting for different forms of violating the instrumental variable assumptions. The success of \texttt{TSCI} requires the instrumental variable's effecZijian Guo, Mengchu Zheng, Peter Bühlmann
  • 09
    Optimising Utility Functions in Multi-Objective Markov Decision Processes
    Multi-Objective Markov Decision Processes (MOMDPs) are among the most prevalent formal frameworks for addressing sequential decision-making problems involving multiple, potentially conflicting objectives. In most MOMDP approaches, a utility function is employed to aggregate these objectives into a single scalar criterion that encodes user preferences. Despite its widespread adoption, the theoretical foundations of MOMDPs remain incomplete in two main respects: first, there is no general characteManel Rodriguez-Soto
  • 10
    Bayesian Transfer Learning for Artificially Intelligent Geospatial Systems: A Predictive Stacking Approach
    Building artificially intelligent geospatial systems requires rapid delivery of spatial data analysis on massive scales with minimal human intervention. Depending on their intended use, learning about underlying spatial processes can also involve model assessment and uncertainty quantification. We devise transfer learning frameworks for deployment in artificially intelligent systems, where a massive data set is split into smaller data sets that stream into the analytical framework to propagate lLuca Presicce, Sudipto Banerjee
  • 11
    Symmetric Rank-k Methods
    This paper proposes a novel class of block quasi-Newton methods for convex optimization which we call symmetric rank-$k$ (SR-$k$) methods. Each iteration of SR-$k$ incorporates the curvature information with $k$ Hessian-vector products achieved from the greedy or random strategy. We prove that SR-$k$ methods have the local superlinear convergence rate of $\mathcal{O}\big((1-k/d)^{t(t-1)/2}\big)$ for minimizing smooth and strongly convex functions, where $d$ is the problem dimension and $t$ is thChengchang Liu, Cheng chen, Luo Luo
  • 12
    From Zipf's Law to Neural Scaling through Heaps' Law and Hilberg's Hypothesis
    We inspect the deductive connection between the neural scaling law and Zipf's law--two statements discussed in machine learning and quantitative linguistics. The neural scaling law describes how the cross entropy rate of a foundation model--such as a large language model--changes with respect to the amount of training tokens, parameters, and compute. By contrast, Zipf's law posits that the distribution of tokens exhibits a power law tail. Whereas similar claims have been made in more specific seŁukasz Dębowski
  • 13
    Efficient Inference under Label Shift in Unsupervised Domain Adaptation
    In many real-world applications, researchers aim to deploy models trained in a source domain to a target domain, where obtaining labeled data is often expensive, time-consuming, or even infeasible. While most existing literature assumes that the source and target data follow the same joint distribution, distribution shifts are common in practice. This paper considers a particular type of distribution shift, label shift, and develops an efficient inference procedure for general parameters charactSeong-ho Lee, Yanyuan Ma, Jiwei Zhao
  • 14
    Adaptive Algorithms for Infinitely Many-Armed Bandits: A Unified Framework
    We consider a bandit problem where the budget is smaller than the number of arms, which may be infinite. In this regime, the usual objective in the literature is to minimize simple regret. To analyze broad classes of distributions with potentially unbounded support, where simple regret may not be well-defined, we take a slightly different approach and seek to maximize the expected simple reward of the recommended arm, providing anytime guarantees. To that end, we introduce a distribution-free alEmmanuel Pilliat
  • 15
    Gradient Estimation for Mixture Variational Inference
    Mixture distributions are expressive variational families for black-box VI, but their discrete component choices complicate gradient estimation. We systematize reparameterization-based estimators for mixtures in a common notation, giving self-contained derivations and extending several to new settings. In particular, we provide an elementary derivation of a single-sample post-stratified estimator---previously derived via transport equations---and prove a variance reduction relative to simple ranJavier Burroni, Daniel Sheldon
  • 16
    torchsom: The Reference PyTorch Library for Self-Organizing Maps
    This paper introduces torchsom, an open-source Python library that provides a reference implementation of the Self-Organizing Map (SOM) in PyTorch. This package offers three main features: (i) dimensionality reduction, (ii) clustering, and (iii) friendly data visualization. It relies on a PyTorch backend, enabling (i) fast and efficient training of SOMs through GPU acceleration, and (ii) easy and scalable integration with the PyTorch ecosystem. torchsom also follows the scikit-learn API for easeLouis Berthier, Ahmed Shokry, Maxime Moreaud, Guillaume Ramelet, Eric Moulines
  • 17
    Pointwise Confidence Estimation in the Non-linear $\ell^2$-regularized Least Squares
    We consider a high-probability non-asymptotic confidence estimation in the $\ell^2$-regularized non-linear least-squares setting with fixed design. In particular, we study confidence estimation for local minimizers of the regularized training loss. We show a pointwise confidence bound, meaning that it holds for the prediction on any given fixed test input $x$. Importantly, the proposed confidence bound scales with similarity of the test input to the training data in the implicit feature space ofIlja Kuzborskij, Yasin Abbasi Yadkori
  • 18
    Safe Learning Under Irreversible Dynamics via Asking for Help
    Most learning algorithms with formal regret guarantees essentially rely on trying all possible behaviors, which is problematic when some errors cannot be recovered from. Instead, we allow the learning agent to ask for help from a mentor and to transfer knowledge between similar states. We show that this combination enables the agent to learn both safely and effectively. Under standard online learning assumptions, we provide an algorithm whose regret and number of mentor queries are both sublineaBenjamin Plaut, Juan Liévano-Karim, Hanlin Zhu, Stuart Russell
  • 19
    AgentPEN: A Prediction-Explanation Network for Sequential Stock Movement via LLMs and Recurrent Generation
    The importance of explainability in stock prediction is increasingly recognized, especially for audit and regulatory purposes. Meanwhile, financial news corpora are often key drivers behind stock price fluctuations. However, the raw news data obtained is usually highly noisy, has a highly variable scope of influence in time and space, and is not precisely synchronized with stock price data. In this paper, we propose a prediction-explanation network called AgentPEN, which can provide clear explanShuqi Li, Mengyao Guo, Yunzhong Zheng, Siqi Li, Xin Gao, Rui Yan
  • 20
    Feedback-Enhanced Online Multiple Testing with Applications to Conformal Selection
    This work studies online multiple testing with feedback, where decisions are made sequentially, and the true state of the hypothesis is revealed after decisions are made, either instantly or with a delay, and under either full or bandit feedback. We propose Generalized alpha-investing with feedback (GAIF) along with its adaptive variants, a feedback-enhanced framework that dynamically adjusts thresholds using revealed outcomes, ensuring finite-sample false discovery rate (FDR)/marginal FDR (mFDRLin Lu, Yuyang Huo, Haojie Ren, Zhaojun Wang, Changliang Zou
  • 21
    Ehrenfeucht-Haussler Rank and Chain of Thought
    The notion of rank of a Boolean function has been a cornerstone in PAC learning, enabling quasipolynomial-time learning algorithms for polynomial-size decision trees. We present a novel characterization of rank, grounded in the well-known Transformer architecture. We show that the rank of a function $f$ corresponds to the minimum number of Chain of Thought (CoT) iterations required by a single-layer Transformer with hard attention to compute $f$. Based on this characterization, we establish tighPablo Barceló, Alexander Kozachinskiy, Tomasz Steifer
  • 22
    scikit-activeml: A Comprehensive and User-Friendly Active Learning Library
    scikit-activeml is a user-friendly open-source Python library for active learning on top of scikit-learn. Included are implementations of a large collection of query strategies, models, and visualization tools in pool- and stream-based active learning for classification or regression tasks with single or multiple annotators. The flexible design of the active learning cycle enables individual adaptations to a variety of learning scenarios. Our source code with comprehensive documentation is availMarek Herde, Minh Tuan Pham, Daniel Kottke, Alexander Benz, Lukas Lührs, Pascal Mergard, Christoph Sandrock, Jiaying Cheng, Atal Roghman, Mehmet Müjde, Lukas Rauch, Bernhard Sick
  • 23
    Locally Private Estimation with Public Features
    We initiate the study of locally differentially private (LDP) learning with public features. We define semi-feature LDP, where some features are publicly available while the remaining ones, along with the label, require protection under local differential privacy. Under semi-feature LDP, we consider three fundamental estimation problems: non-parametric density estimation, classification, and regression. Given the smoothness assumption, we show that the minimax convergence rate is significantly iYuheng Ma, Hanfang Yang, Ke Jia
  • 24
    Dimension Reduction for Derivative-Informed Operator Learning: An Analysis of Approximation Errors
    We study the derivative-informed learning of nonlinear operators between infinite-dimensional Hilbert spaces. Such operators can arise as solution maps of partial differential equations, and their approximation by accurate surrogate models can accelerate simulation-intensive tasks of scientific and engineering interest, including inference, control, and uncertainty quantification. Since efficiently performing such tasks often requires an accurate representation of the operator's derivatives, weDingcheng Luo, Thomas O'Leary-Roseberry, Peng Chen, Omar Ghattas
  • 25
    Domain Adaptation Targeting Heterogeneous and Imbalanced Subgroups
    Domain adaptation enables generalizable and efficient data-driven research. However, existing work has largely focused on domain adaptation for some intrinsically homogeneous target cohort, overlooking inherent heterogeneity within the target, which can exacerbate biases and unfairness in the presence of subgroups with imbalanced sample sizes. We develop a novel domain adaptation framework that addresses a more complicated target dataset that consists of heterogeneous and data-sparse subgroups aDoudou Zhou, Mengyan Li, Yun Wang, Tianxi Cai, Molei Liu
  • 26
    On the Effectiveness of the z-Transform Method in Quadratic Optimization
    The z-transform of a sequence is a classical tool used in signal processing, control theory, computer science, and electrical engineering. It allows one to study sequences from their generating functions, with many operations that can be equivalently defined on the original sequence and its z-transform. In particular, the z-transform method focuses on asymptotic behaviors and allows the use of Taylor expansions. We present a sequence of results of increasing significance and difficulty for lineaFrancis Bach
  • 27
    Breaking the Curse of Dimensionality: Diffusion Models Efficiently Learn Low-Dimensional Distributions
    Despite their empirical success across a wide range of generative tasks, the fundamental principles underlying the ability of diffusion models to learn data distributions are poorly understood. In this work, we develop a new mathematical framework that explains how diffusion models can effectively learn low-dimensional distributions from a finite number of training samples without suffering from the curse of dimensionality. Specifically, motivated by the intrinsic low-dimensional structure of imPeng Wang, Huijie Zhang, Zekai Zhang, Siyi Chen, Yi Ma, Qing Qu
  • 28
    Consistency of Augmentation Graph and Network Approximability in Contrastive Learning
    Contrastive learning leverages data augmentation to develop feature representation without relying on large labeled data sets. However, despite its empirical success, the theoretical foundations of contrastive learning remain incomplete, with many essential guarantees left unaddressed, particularly the realizability assumption concerning neural approximability of an optimal spectral contrastive loss solution. In this work, we overcome these limitations by analyzing pointwise and spectral consistChenghui Li, A. Martina Neuman
  • 29
    Solving Nonlinear PDEs with Sparse Radial Basis Function Networks
    We propose a novel framework for solving nonlinear PDEs using sparse radial basis function (RBF) networks. Sparsity-promoting regularization is employed to prevent over-parameterization and reduce redundant features. This work is motivated by longstanding challenges in traditional RBF collocation methods, along with the limitations of physics-informed neural networks (PINNs) and Gaussian process (GP) approaches, aiming to blend their respective strengths in a unified framework. The theoretical fZihan Shao, Konstantin Pieper, Xiaochuan Tian
  • 30
    Viscosity Convergence Analysis for Deep Q-Networks
    Deep Q-Networks (DQNs) and related residual neural architectures are increasingly used for continuous-time reinforcement learning (CTRL), where optimal value functions solve fully nonlinear second-order Hamilton--Jacobi--Bellman (HJB) equations and may be non-smooth. In this regime, convergence should be analyzed in the viscosity-solution framework. We study a spatially-coupled monotone ResNet architecture whose one-step operator is a non-negative local aggregation, designed to satisfy monotonicQian Qi
  • 31
    Clustering and Pruning in Causal Data Fusion
    Data fusion, the process of combining observational and experimental data, can enable the identification of causal effects that would otherwise remain non-identifiable. Although identification algorithms have been developed for specific scenarios, do-calculus remains the only general-purpose tool for causal data fusion, particularly when variables are present in some data sources but not others. However, approaches based on do-calculus may encounter computational challenges as the number of variOtto Tabell, Santtu Tikka, Juha Karvanen
  • 32
    Leakage and Interpretability in Concept-Based Models
    Concept-based Models aim to improve interpretability by predicting high-level intermediate concepts, representing a promising approach for deployment in high-risk scenarios. However, they are known to suffer from information leakage, whereby models exploit unintended information encoded within the learned concepts. We introduce an information-theoretic framework to rigorously characterise and quantify leakage, and define two complementary measures: the concepts-task leakage (CTL) and interconcepEnrico Parisini, Tapabrata Chakraborti, Chris Harbron, Ben D. MacArthur, Christopher R.S. Banerji
  • 33
    Resilience Beyond Stationary Client Unavailability: Unlocking Efficient and Unbiased Federated Learning
    Due to resource constraints or external and internal uncertainties, clients in real-world federated learning systems are often intermittently available edge devices. In highly dynamic environments, the parameter server lacks prior real-time knowledge of clients' availability, making it challenging to adapt traditional federated learning algorithms to be resilient to uncertainties in client availability. If not carefully addressed, complex client availability can introduce significant bias, potenMing Xiang, Stratis Ioannidis, Edmund Yeh, Carlee Joe-Wong, Lili Su
  • 34
    Incorporating external data for analyzing randomized clinical trials: A transfer learning approach
    Randomized clinical trials are the gold standard for analyzing treatment effects. However, increasing costs and ethical concerns may limit trial recruitment, resulting in insufficient sample sizes and potentially invalid inference. Incorporating external trial data with similar characteristics (treatments, diseases, biomarkers, etc.) into the analysis appears promising for addressing these issues. Transfer learning, which in our context utilizes external trials as the source domain and current tYujia Gu, Hanzhong Liu, Wei Ma
  • 35
    A Provably Convergent Plug-and-Play Framework for Stochastic Bilevel Optimization
    Bilevel optimization has recently attracted significant attention in machine learning due to its wide range of applications and advanced hierarchical optimization capabilities. In this paper, we propose a plug-and-play framework, named PnPBO, for developing and analyzing stochastic bilevel optimization methods. This framework integrates both modern unbiased and biased stochastic estimators into the single-loop bilevel optimization framework introduced in Dagréou et al. (2022), with several improTianshu Chu, Dachuan Xu, Wei Yao, Chengming Yu, Jin Zhang
  • 36
    Particle Filter for Bayesian Inference on Privatized Data
    Differential privacy is a probabilistic framework that protects privacy while preserving data utility. To protect the privacy of the individuals in the data set, differential privacy requires adding a precise amount of noise to a statistic of interest; however, this noise addition alters the resulting sampling distribution, making statistical inference challenging. One of the main differential privacy goals in Bayesian analysis is to make statistical inference based on the private posterior distYu-Wei Chen, Pranav Sanghi, Jordan Awan
  • 37
    Test-time regression: a unifying framework for designing sequence models with associative memory
    Sequence models lie at the heart of modern deep learning. However, rapid advancements have produced a diversity of seemingly unrelated architectures, such as Transformers and recurrent alternatives. In this paper, we introduce a unifying framework to understand and derive these sequence models, inspired by the empirical importance of associative recall, the capability to retrieve contextually relevant tokens. We formalize associative recall as a two-step process, memorization and retrieval, castKe Alexander Wang, Jiaxin Shi, Emily B. Fox
  • 38
    Nonparametric Spectral Density Estimation using Interactive Mechanisms under Local Differential Privacy
    We study the problem of estimating the spectral density of a centered stationary Gaussian time series under local differential privacy constraints. Specifically, we propose new interactive privacy mechanisms for three tasks: recovering a single covariance coefficient, recovering the spectral density at a fixed frequency, and global recovery. Our approach achieves faster rates through a two-stage process: we first apply the Laplace mechanism to the truncated value, and then use the resulting privCristina Butucea, Karolina Klockmann, Tatyana Krivobokova
  • 39
    Sublinear Variational Optimization of Gaussian Mixture Models with Millions to Billions of Parameters
    Gaussian Mixture Models (GMMs) range among the most frequently used models in machine learning. However, training large, general GMMs becomes computationally prohibitive for data sets that have many data points $N$ of high-dimensionality $D$. For GMMs with arbitrary covariances, we here derive a highly efficient variational approximation, which is then integrated with mixtures of factor analyzers (MFAs). For GMMs with $C$ components, our proposed algorithm substantially reduces runtime complexitSebastian Salwig, Till Kahlke, Florian Hirschberger, Dennis Forster, Jörg Lücke
  • 40
    Statistical Inference for High-dimensional Partially Linear Models via Debiased Rank Lasso
    This paper aims to develop tuning-free and robust regularized methods for partially linear models based on partial residual methods. In order to preserve the near-oracle rate of rank Lasso, we construct a new estimator via a data-splitting procedure and name the resulting estimator as data-splitting (DS) rank Lasso, which is based on partial prediction residuals. Thus, the proposed estimation procedure is distinguished from the traditional partial residual based methods, which are based on the mSongshan Yang, Delin Zhao, Runze Li
  • 41
    Identifiability of the Instrumental Variable Model with the Treatment and Outcome Missing Not at Random
    Under the instrumental variable model, we can identify the local average treatment effect, also known as the complier average causal effect (CACE). In practice, however, the treatment and outcome are often missing. When they are missing not at random (MNAR), the underlying data distribution cannot be recovered, so the CACE is generally not identifiable without further assumptions. We study the conditions under which the CACE remains identifiable when data are MNAR. Searching exhaustively over miShuozhi Zuo, Peng Ding, Fan Yang
  • 42
    Deep Neural Expected Shortfall Regression with Tail-Robustness
    Expected shortfall (ES), also known as conditional value-at-risk, is a widely recognized risk measure that complements value-at-risk by capturing tail-related risks more effectively. Compared with quantile regression, which has been extensively developed and applied across disciplines, ES regression remains in its early stage, partly because the traditional empirical risk minimization framework is not directly applicable. In this paper, we develop a nonparametric framework for expected shortfallMyeonghun Yu, Kean Ming Tan, Huixia Judy Wang, Wen-Xin Zhou
  • 43
    Optimal Convergence Rates for Neural Operators
    We introduce the neural tangent kernel (NTK) regime for two-layer neural operators and analyze their generalization properties. For early-stopped gradient descent (GD), we derive fast convergence rates that are known to be minimax optimal within the framework of non-parametric regression in reproducing kernel Hilbert spaces (RKHS). We provide bounds on the number of hidden neurons and the number of second-stage samples necessary for generalization. To justify our NTK regime, we additionally showMike Nguyen, Nicole Mücke
  • 44
    Have ASkotch: A Neat Solution for Large-Scale Kernel Ridge Regression
    Kernel ridge regression (KRR) is a fundamental computational tool, appearing in problems that range from computational chemistry to health analytics, with a particular interest due to its starring role in Gaussian process regression. However, full KRR solvers are challenging to scale to large datasets: both direct (e.g., Cholesky decomposition) and iterative methods (e.g., PCG) incur prohibitive computational and storage costs. The standard approach to scale KRR to large datasets chooses a set oPratik Rathore, Zachary Frangella, Jiaming Yang, Michał Dereziński, Madeleine Udell
  • 45
    Pairwise Comparisons without Stochastic Transitivity: Model, Theory and Applications
    Most statistical models for pairwise comparisons, including the Bradley-Terry (BT) and Thurstone models and many extensions, make a relatively strong assumption of stochastic transitivity. This assumption imposes the existence of an unobserved global ranking among all the players/teams/items and monotone constraints on the comparison probabilities implied by the global ranking. However, the stochastic transitivity assumption does not hold in many real-world scenarios of pairwise comparisons, espSze Ming Lee, Yunxiao Chen
  • 46
    Canonical Correlation Analysis as Reduced Rank Regression in High Dimensions
    Canonical correlation analysis is a widespread technique for discovering linear relationships between two sets of variables. In high dimensions, however, standard estimates of the canonical directions cease to be consistent without assuming further structure. In this setting, a possible solution consists in leveraging the presumed sparsity of the solution: only a subset of the covariates span the canonical directions. While the last decade has seen a proliferation of sparse canonical correlationClaire Donnat, Elena Tuzhilina
  • 47
    Bayesian Level Set Clustering
    Classically, Bayesian clustering interprets each component of a mixture model as a cluster. The inferred clustering posterior is highly sensitive to any inaccuracies in the kernel within each component. As this kernel is made more flexible, problems arise in identifying the underlying clusters in the data. To address this pitfall, this article proposes a fundamentally different approach to Bayesian clustering that decouples the problems of clustering and flexible modeling of the data density f.David Buch, Miheer Dewaskar, David B. Dunson
  • 48
    Differentially Private Synthetic Data Generation for Relational Databases
    Existing differentially private (DP) synthetic data generation mechanisms typically assume a single-source table. In practice, data is often distributed across multiple tables with relationships across tables. In this paper, we introduce the first-of-its-kind algorithm that can be combined with any existing DP mechanisms to generate synthetic relational databases. Our algorithm iteratively refines the relationship between individual synthetic tables to minimize their approximation errors in termKaveh Alim, Hao Wang, Ojas Gulati, Akash Srivastava, Navid Azizan
  • 49
    A Neural Network Approach to Learning Solutions of a Class of Elliptic Variational Inequalities
    We develop a weak adversarial approach to solving obstacle problems using neural networks. By employing (generalised) regularised gap functions and their properties we rewrite the obstacle problem (which is an elliptic variational inequality) as a minmax problem, providing a natural formulation amenable to learning. Our approach, in contrast to much of the literature, does not require the elliptic operator to be symmetric. We provide an error analysis for suitable discretisations of the continuoAmal Alphonse, Michael Hintermüller, Alexander Kister, Chin Hang Lun, Clemens Sirotenko
  • 50
    Impatient Bandits: Optimizing for the Long-Term Without Delay
    Increasingly, recommender systems are tasked with improving users' long-term satisfaction. In this context, we study a content exploration task, which we formalize as a bandit problem with delayed rewards. There is an apparent trade-off in choosing the learning signal: waiting for the full reward to become available might take several weeks, slowing the rate of learning, whereas using short-term proxy rewards reflects the actual long-term goal only imperfectly. First, we develop a predictive modKelly W. Zhang, Thomas Baldwin-McDonald, Kamil Ciosek, Lucas Maystre, Daniel Russo
  • 51
    Conditional Regression for the Nonlinear Single-Variable Model
    Regressing a function $F$ on $\mathbb{R}^d$ without incurring the statistical and computational curse of dimensionality requires exploitable structure. Compositional models $F=f\circ g$ in which $g$ has a low-dimensional range include classical single- and multi-index models as well as certain neural networks; while the case of linear $g$ is well understood, substantially less is known for nonlinear $g$. We study the model $F(X)=f(\Pi_\gamma X)$, where $\Pi_\gamma$ is the closest-point coordinatYantao Wu, Mauro Maggioni
共 205 条热搜查看完整榜单
共 20 条热搜查看完整榜单
  • 01
    #502 – Psychiatry, Insane Asylums, Mental Illness, ECT, Lobotomies, Freud & Jung
    Andrew Scull is a historian of psychiatry. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep502-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/andrew-scull-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hiring Other – oth
  • 02
    #501 – DHH: Future of Programming, AI, Agentic Engineering, Vibe Coding & Linux
    DHH is the creator of Ruby on Rails, Omarchy Linux, CTO of 37signals, and a racecar driver. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep501-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/dhh-2-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team:
  • 03
    #500 – Khabib Nurmagomedov: Dagestan, MMA, UFC, Islam, Conor, Fedor & Football
    Khabib Nurmagomedov is one of the greatest fighters of all time, who retired from the UFC undefeated with a perfect 29-0 record. We did this conversation entirely in Russian. It’s translated and dubbed into English. Both language audio tracks (and subtitles) are available on YouTube. Here, only the English audio is available. We worked hard to make it enjoyable to listen to, by carefully dubbing the translation using voice-cloning, as we’ve done for previous foreign-language podcasts. Thank you
  • 04
    #499 – Gary Gallagher: American Civil War, Slavery, Lincoln, Grant & Lee
    Gary Gallagher is a historian of the American Civil War. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep499-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/gary-gallagher-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hi
  • 05
    #498 – Anthony Kaldellis: Roman Empire, Byzantine Empire, Rise & Fall of Empires
    Anthony Kaldellis is a historian of the Roman Empire and author of “The New Roman Empire”, a comprehensive history of the Byzantine Empire (Eastern Roman Empire). Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep498-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/anthony-kaldellis-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – subm
  • 06
    #497 – Biggest Mysteries in Physics: Antimatter, Dark Energy & ToE – Don Lincoln
    Don Lincoln is a particle physicist at Fermilab who has spent decades working at the frontiers of high energy physics. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep497-sc See below for timestamps, and to give feedback, submit questions, contact Lex, etc. CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hiring Other
  • 07
    #496 – FFmpeg: The Incredible Technology Behind Video on the Internet
    Jean-Baptiste Kempf is lead developer of VLC and president of VideoLAN. Kieran Kunhya is a longtime FFmpeg contributor, codec engineer, and the person behind the now-infamous FFmpeg account on X. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep496-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/ffmpeg-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman
  • 08
    #495 – Vikings, Ragnar, Berserkers, Valhalla & the Warriors of the Viking Age
    Lars Brownworth is a historian, teacher, podcaster, and author specializing in Viking history, medieval Europe, and the Byzantine Empire. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep495-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/lars-brownworth-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or cal
  • 09
    #494 – Jensen Huang: NVIDIA – The $4 Trillion Company & the AI Revolution
    Jensen Huang is the co-founder and CEO of NVIDIA, the world’s most valuable company and the engine powering the AI computing revolution. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep494-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/jensen-huang-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in
  • 10
    #493 – Jeff Kaplan: World of Warcraft, Overwatch, Blizzard, and Future of Gaming
    Jeff Kaplan is a legendary Blizzard game designer of World of Warcraft and Overwatch, now preparing to launch a new game, The Legend of California, from his new studio Kintsugiyama – available to wishlist on Steam today, with alpha later in March. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep493-sc See below for timestamps, and to give feedback, submit questions, contact Lex, etc. CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey A
  • 11
    #492 – Rick Beato: Greatest Guitarists of All Time, History & Future of Music
    Rick Beato is a music educator, interviewer, producer, songwriter, and a true multi-instrument musician, playing guitar, bass, cello & piano. His incredible YouTube channel celebrates great musicians & musical ideas, and helps millions of people fall in love with great music all over again. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep492-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://l
  • 12
    #491 – OpenClaw: The Viral AI Agent that Broke the Internet – Peter Steinberger
    Peter Steinberger is the creator of OpenClaw, an open-source AI agent framework that’s the fastest-growing project in GitHub history. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep491-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/peter-steinberger-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-
  • 13
    #490 – State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI
    Nathan Lambert and Sebastian Raschka are machine learning researchers, engineers, and educators. Nathan is the post-training lead at the Allen Institute for AI (Ai2) and the author of The RLHF Book. Sebastian Raschka is the author of Build a Large Language Model (From Scratch) and Build a Reasoning Model (From Scratch). Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep490-sc See below for timestamps, transcript, and to give feedback, submit questions, contact L
  • 14
    #489 – Paul Rosolie: Uncontacted Tribes in the Amazon Jungle
    Paul Rosolie is a naturalist, explorer, author of a new book titled Junglekeeper, and is someone who has dedicated his life to protecting the Amazon rainforest. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep489-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/paul-rosolie-3-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit qu
  • 15
    #488 – Infinity, Paradoxes that Broke Mathematics, Gödel Incompleteness & the Multiverse – Joel David Hamkins
    Joel David Hamkins is a mathematician and philosopher specializing in set theory, the foundations of mathematics, and the nature of infinity, and he’s the #1 highest-rated user on MathOverflow. He is also the author of several books, including Proof and the Art of Mathematics and Lectures on the Philosophy of Mathematics. And he has a great blog called Infinitely More. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep488-sc See below for timestamps, transcript,
  • 16
    #487 – Irving Finkel: Deciphering Secrets of Ancient Civilizations & Flood Myths
    Irving Finkel is a scholar of ancient languages and a longtime curator at the British Museum, renowned for his expertise in Mesopotamian history and cuneiform writing. He specializes in reading and interpreting cuneiform inscriptions, including tablets from Sumerian, Akkadian, Babylonian, and Assyrian contexts. He became widely known for studying a tablet with a Mesopotamian flood story that predates the biblical Noah narrative, which he presented in his book “The Ark Before Noah” and in a docum
  • 17
    #486 – Michael Levin: Hidden Reality of Alien Intelligence & Biological Life
    Michael Levin is a biologist at Tufts University working on novel ways to understand and control complex pattern formation in biological systems. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep486-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/michael-levin-2-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, video
  • 18
    #485 – David Kirtley: Nuclear Fusion, Plasma Physics, and the Future of Energy
    David Kirtley is a nuclear fusion engineer and CEO of Helion Energy, a company working on building the world’s first commercial fusion power plant by 2028. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep485-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/david-kirtley-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit question
  • 19
    #484 – Dan Houser: GTA, Red Dead Redemption, Rockstar, Absurd & Future of Gaming
    Dan Houser is co-founder of Rockstar Games and is a legendary creative mind behind Grand Theft Auto (GTA) and Red Dead Redemption series of video games. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep484-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/dan-houser-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, vid
  • 20
    #483 – Julia Shaw: Criminal Psychology of Murder, Serial Killers, Memory & Sex
    Julia Shaw is a criminal psychologist and author who in her books explores human nature, including psychopathy, violent crime, the psychology of evil, police interrogation, false memory manipulation, deception detection, and human sexuality. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep483-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/julia-shaw-transcript CONTACT LEX: F
  • 21
    #482 – Pavel Durov: Telegram, Freedom, Censorship, Money, Power & Human Nature
    Pavel Durov is the founder and CEO of Telegram. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep482-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/pavel-durov-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hiring Other –
  • 22
    #481 – Norman Ohler: Hitler, Nazis, Drugs, WW2, Blitzkrieg, LSD, MKUltra & CIA
    Norman Ohler is a historian and author of “Blitzed: Drugs in the Third Reich,” a book that investigates the role of psychoactive drugs, particularly stimulants such as methamphetamine, in the military history of World War II. It is a book that two legendary historians Ian Kershaw and Antony Beevor give very high praise for its depth of research. Norman also wrote “Tripped: Nazi Germany, the CIA, and the Dawn of the Psychedelic Age”, and he is working on a new book “Stoned Sapiens” looking at the
  • 23
    #480 – Dave Hone: T-Rex, Dinosaurs, Extinction, Evolution, and Jurassic Park
    Dave Hone is a paleontologist, expert on dinosaurs, co-host of the Terrible Lizards podcast, and author of numerous scientific papers and books on the behavior and ecology of dinosaurs. He lectures at Queen Mary University of London on topics of Ecology, Zoology, Biology, and Evolution. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep480-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfr
  • 24
    #479 – Dave Plummer: Programming, Autism, and Old-School Microsoft Stories
    Dave Plummer is a programmer, former Microsoft software engineer (Windows 95, NT, XP), creator of Task Manager, author of two books on autism, and host of the Dave’s Garage YouTube channel, where he shares stories from his career, insights on software development, and deep dives into technology. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep479-sc See below for timestamps, and to give feedback, submit questions, contact Lex, etc. CONTACT LEX: Feedback – give
  • 25
    #478 – Scott Horton: The Case Against War and the Military Industrial Complex
    Scott Horton is the director of the Libertarian Institute, editorial director of Antiwar.com, host of The Scott Horton Show, co-host of Provoked, and for the past three decades a staunch critic of U.S. military interventionism. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep478-sc See below for timestamps, and to give feedback, submit questions, contact Lex, etc. CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit question
  • 26
    #477 – Keyu Jin: China’s Economy, Tariffs, Trade, Trump, Communism & Capitalism
    Keyu Jin is an economist specializing in China’s economy, international macroeconomics, global trade imbalances, and financial policy. She is the author of The New China Playbook: Beyond Socialism and Capitalism. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep477-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/keyu-jin-transcript CONTACT LEX: Feedback – give feedback to Lex:
  • 27
    #476 – Jack Weatherford: Genghis Khan and the Mongol Empire
    Jack Weatherford is an anthropologist and historian specializing in Genghis Khan and the Mongol Empire. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep476-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/jack-weatherford-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama
  • 28
    #475 – Demis Hassabis: Future of AI, Simulating Reality, Physics and Video Games
    Demis Hassabis is the CEO of Google DeepMind and Nobel Prize winner for his groundbreaking work in protein structure prediction using AI. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep475-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/demis-hassabis-2-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or ca
  • 29
    #474 – DHH: Future of Programming, AI, Ruby on Rails, Productivity & Parenting
    David Heinemeier Hansson (aka DHH) is a legendary programmer, creator of Ruby on Rails, co-owner & CTO of 37signals that created Basecamp, HEY, & ONCE, and is a NYT-best-selling author (with Jason Fried) of 4 books: REWORK, REMOTE, Getting Real, and It Doesn’t Have To Be Crazy At Work. He is also a race car driver, including a class-winning performance at the 24 hour Le Mans race. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep474-sc See below for timestamps,
  • 30
    #473 – Iran War Debate: Nuclear Weapons, Trump, Peace, Power & the Middle East
    Debate on Iran war between Scott Horton and Mark Dubowitz. Scott Horton is the author and director of the Libertarian Institute, editorial director of Antiwar.com, host of The Scott Horton Show, and for the past three decades, a staunch critic of U.S. foreign policy and military interventionism. Mark Dubowitz is the chief executive of the Foundation for Defense of Democracies, host of the Iran Breakdown podcast, and a leading expert on Iran and its nuclear program for over 20 years. This debate
  • 31
    #472 – Terence Tao: Hardest Problems in Mathematics, Physics & the Future of AI
    Terence Tao is widely considered to be one of the greatest mathematicians in history. He won the Fields Medal and the Breakthrough Prize in Mathematics, and has contributed to a wide range of fields from fluid dynamics with Navier-Stokes equations to mathematical physics & quantum mechanics, prime numbers & analytics number theory, harmonic analysis, compressed sensing, random matrix theory, combinatorics, and progress on many of the hardest problems in the history of mathematics. Thank you for
  • 32
    #471 – Sundar Pichai: CEO of Google and Alphabet
    Sundar Pichai is CEO of Google and Alphabet. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep471-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/sundar-pichai-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hiring Other –
  • 33
    #470 – James Holland: World War II, Hitler, Churchill, Stalin & Biggest Battles
    James Holland is a historian specializing in World War II. He hosts a podcast called WW2 Pod: We Have Ways of Making You Talk. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep470-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/james-holland-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https:/
  • 34
    #469 – Oliver Anthony: Country Music, Blue-Collar America, Fame, Money, and Pain
    Oliver Anthony is singer-songwriter who first gained worldwide fame with his viral hit Rich Men North of Richmond. He became a voice for many who are voiceless, with many of his songs speaking to the struggle of the working class in modern American life. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep469-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/oliver-anthony-transcri
  • 35
    #468 – Janna Levin: Black Holes, Wormholes, Aliens, Paradoxes & Extra Dimensions
    Janna Levin is a theoretical physicist and cosmologist specializing in black holes, cosmology of extra dimensions, topology of the universe, and gravitational waves. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep468-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/janna-levin-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit
  • 36
    #467 – Tim Sweeney: Fortnite, Unreal Engine, and the Future of Gaming
    Tim Sweeney is a legendary video game programmer, founder and CEO of Epic Games that created the Unreal Engine, Fortnite, Gears of War, Unreal Tournament, and many other groundbreaking and influential video games. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep467-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/tim-sweeney-transcript CONTACT LEX: Feedback – give feedback to
  • 37
    #466 – Jeffrey Wasserstrom: China, Xi Jinping, Trade War, Taiwan, Hong Kong, Mao
    Jeffrey Wasserstrom is a historian of modern China. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep466-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/jeffrey-wasserstrom-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team: https://lexfridman.com/hi
  • 38
    #465 – Robert Rodriguez: Sin City, Desperado, El Mariachi, Alita, and Filmmaking
    Robert Rodriguez is a legendary filmmaker and creator of Sin City, El Mariachi, Desperado, Spy Kids, Machete, From Dusk Till Dawn, Alita: Battle Angel, The Faculty, and his newest venture Brass Knuckle Films. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep465-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/robert-rodriguez-transcript CONTACT LEX: Feedback – give feedback to
  • 39
    #464 – Dave Smith: Israel, Ukraine, Epstein, Mossad, Conspiracies & Antisemitism
    Dave Smith is a comedian, libertarian, political commentator, and the host of Part of the Problem podcast. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep464-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/dave-smith-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hir
  • 40
    #463 – Douglas Murray: Putin, Zelenskyy, Trump, Israel, Netanyahu, Hamas & Gaza
    Douglas Murray is the author of On Democracies and Death Cults, The War on The West, and The Madness of Crowds. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep463-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/douglas-murray-2-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.
  • 41
    #462 – Ezra Klein and Derek Thompson: Politics, Trump, AOC, Elon & DOGE
    Ezra Klein is one of the most influential voices representing the left-wing of American politics. He is a columnist for the NY Times and host of The Ezra Klein Show. Derek Thompson is a writer at The Atlantic and host of the Plain English podcast. Together they have written a new book titled Abundance that lays out a set of ideas for the future of the Democratic party. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep462-sc See below for timestamps, transcript,
  • 42
    #461 – ThePrimeagen: Programming, AI, ADHD, Productivity, Addiction, and God
    ThePrimeagen (aka Michael Paulson) is a programmer who has educated, entertained, and inspired millions of people to build software and have fun doing it. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep461-sc See below for timestamps, and to give feedback, submit questions, contact Lex, etc. CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring – join our team:
  • 43
    #460 – Narendra Modi: Prime Minister of India – Power, Democracy, War & Peace
    Narendra Modi is the Prime Minister of India. On YouTube this episode is available in English, Hindi, Russian (and soon other languages). Captions and voice-over audio tracks are provided (for the main episode video on YouTube) in English, Hindi, Russian, and the original mixed-language version, with subtitles available in your preferred language. To listen to the original mixed-language version, please select the Hindi (Latin) audio track. The default is English overdub. Thank you for listening
  • 44
    #459 – DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters
    Dylan Patel is the founder of SemiAnalysis, a research & analysis company specializing in semiconductors, GPUs, CPUs, and AI hardware. Nathan Lambert is a research scientist at the Allen Institute for AI (Ai2) and the author of a blog on AI called Interconnects. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep459-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/deepseek-dylan-
  • 45
    #458 – Marc Andreessen: Trump, Power, Tech, AI, Immigration & Future of America
    Marc Andreessen is an entrepreneur, investor, co-creator of Mosaic, co-founder of Netscape, and co-founder of the venture capital firm Andreessen Horowitz. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep458-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/marc-andreessen-2-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit ques
  • 46
    #457 – Jennifer Burns: Milton Friedman, Ayn Rand, Economics, Capitalism, Freedom
    Jennifer Burns is a historian of ideas, focusing on the evolution of economic, political, and social ideas in the United States in the 20th century. She wrote two biographies, one on Milton Friedman, and the other on Ayn Rand. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep457-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/jennifer-burns-transcript CONTACT LEX: Feedback – g
  • 47
    #456 – Volodymyr Zelenskyy: Ukraine, War, Peace, Putin, Trump, NATO, and Freedom
    Volodymyr Zelenskyy is the President of Ukraine. On YouTube this episode is available in English, Ukrainian, and Russian. Captions and voice-over audio tracks are provided in English, Ukrainian, Russian, and the original mixed-language version, with subtitles available in your preferred language. To listen to the original mixed language version, please select the English (UK) audio track audio track. The default is English overdub. Thank you for listening ❤ Check out our sponsors: https://lexfri
  • 48
    #455 – Adam Frank: Alien Civilizations and the Search for Extraterrestrial Life
    Adam Frank is an astrophysicist studying star systems and the search for extraterrestrial life and alien civilizations. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep455-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/adam-frank-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridma
  • 49
    #454 – Saagar Enjeti: Trump, MAGA, DOGE, Obama, FDR, JFK, History & Politics
    Saagar Enjeti is a political journalist & commentator, co-host of Breaking Points with Krystal and Saagar and The Realignment Podcast. He is exceptionally well-read, and the books he recommends are always fascinating and eye-opening. You can check out all the books he mentions in this episode here: https://lexfridman.com/saagar-books Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep454-sc See below for timestamps, transcript, and to give feedback, submit questi
  • 50
    #453 – Javier Milei: President of Argentina – Freedom, Economics, and Corruption
    Javier Milei is the President of Argentina. This episode is available in both English and Spanish. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep453-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/javier-milei-transcript CONTACT LEX: Feedback – give feedback to Lex: https://lexfridman.com/survey AMA – submit questions, videos or call-in: https://lexfridman.com/ama Hiring –
  • 51
    #452 – Dario Amodei: Anthropic CEO on Claude, AGI & the Future of AI & Humanity
    Dario Amodei is the CEO of Anthropic, the company that created Claude. Amanda Askell is an AI researcher working on Claude’s character and personality. Chris Olah is an AI researcher working on mechanistic interpretability. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep452-sc See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. Transcript: https://lexfridman.com/dario-amodei-transcript CONTACT LEX: Feedback – give f
共 503 条热搜查看完整榜单
  • 01
    Importing Data using Java Table Functions
    When the analytics screen running on our main operational database got too slow, we moved a year of data into DuckDB on the same server. Getting the data in was the hard part. This is the story of every method we tried, and why a table function written in pure Java is the one we shipped.Guest Author, Geertjan Wielenga, Alex Kasko
  • 02
    Faster String Aggregations with Dimension Tables
    When a query groups on long, repeated strings, move the strings into a small dimension table with sorted, narrow integer keys. Aggregate on the keys, then join the strings back in at the very end. The query works as before, though on small fixed-width integers instead of variable-length text.DuckDB team
  • 03
    Jev and DuckDB: Plain-English Conditions in SQL
    TypeSafe AI released Jev, a model that returns typed answers instead of text, on September 15, 2026. Within ten days, several community extensions let you filter, classify and score DuckDB rows with plain-English conditions. This post covers what Jev is, how the extensions work and the posts and videos about them.Geertjan Wielenga, Gábor Szárnyas
  • 04
    Announcing DuckDB 1.5.6
    Today we are releasing DuckDB 1.5.6 with bugfixes and performance improvements.The DuckDB team
  • 05
    DuckDB and Hugging Face: Querying Datasets Directly
    Hugging Face hosts hundreds of thousands of datasets, and DuckDB can read them directly, exactly where they are and without downloading anything, over the DuckDB `hf://` protocol. This post looks at how the integration works and the scenarios where it works best.The DuckDB team
  • 06
    DuckDB Now Ships inside dbt v2
    dbt v2, which runs on the new Rust-based Fusion engine, is the first dbt release that ships with a built-in DuckDB adapter. This post covers setup, DuckLake and Iceberg catalogs, querying dbt's Parquet metadata with DuckDB, plus other v2 features that matter to DuckDB users, including migrating to dbt v2.Geertjan Wielenga
  • 07
    Persistent Databases in the Browser with DuckDB-Wasm and OPFS
    DuckDB-Wasm can open a persistent database file in the browser's Origin Private File System (OPFS). This post shows how, and when data reaches disk.Carlo Piovesan, Geertjan Wielenga
  • 08
    DuckDB Skills for Claude Code
    The duckdb-skills plugin gives Claude Code a growing number of skills that use the DuckDB CLI to read data files, run queries, convert formats, explore object storage, work with spatial data, search the documentation and recall earlier sessions.The DuckDB team
  • 09
    Try DuckDB v2.0-dev
    DuckDB's development team in Amsterdam has started getting DuckDB v2.0 ready for release in October. If you like shiny new things, try out the alpha releases now and report anything that might not be working as expected!The DuckDB team
  • 10
    DuckLabs to Join AWS, Projects to Remain Open Source
    DuckLabs will join Amazon Web Services (AWS), which is expected to be effective in early September. The projects will remain open-source under the MIT license.Mark Raasveldt and Hannes Mühleisen
共 10 条热搜查看完整榜单
共 40 条热搜查看完整榜单
共 20 条热搜查看完整榜单
  • 01
    What’s the Future for Pure Math Research in the Age of AI?
    Headlines and History The headlines keep coming: such and such an AI system has solved such and such a math problem. And more and more I’m hearing people saying: maybe we don’t need people doing math research anymore; maybe we should just delegate it all to more and more powerful AIs. I must admit that […]Stephen Wolfram
  • 02
    In Memory of My Wife, Elise Cawley (1961–2026), with Thanks for 36 Wonderful Years
    Something terrible just happened. My wife, Elise Cawley, was recovering from heart surgery and had just attended virtually a celebration for one of our children when she had a freak, vast cardiovascular event—and died instantly. We had been together for 36 years. The picture above was taken just hours before she died. In all the […]Stephen Wolfram
  • 03
    Towards a Theory of Bugs: The Ruliology of the Unexpected
    “My Program Did the Wrong Thing!” Bugs are a ubiquitous phenomenon in the software world. And—essentially by definition—each one of them is somehow unique and unexpected. But—particularly given their ubiquity—one might wonder whether there could perhaps be some kind of general “scientific” theory that could be developed about them. My goal here is to explore […]Stephen Wolfram
  • 04
    Launching Version 15 of Wolfram Language & Mathematica: Built-in (Useful) AI & Lots of New Core Functionality
    June 23, 1988 is when we launched Version 1.0 of Mathematica . Today—almost 38 years later—we’re launching Version 15 of what—in recognition of how far it’s expanded beyond “math”—we now call Wolfram Language . It’s an impressive release, with a lot of new core functionality. It might perhaps seem surprising that after 38 years there’d still be more to add. But it’s like the typical arc of intellectual history: the more one’s figured out, the further one can see, and the more one becomes able toStephen Wolfram
  • 05
    Games between Programs: The Ruliology of Competition
    The Basic Setup Whether one’s dealing with biology, economics, politics or a host of other fields, it’s common to encounter situations that can be modeled as involving two agents that repeatedly compete with each other. One imagines that at each step each agent can take one of a certain set of actions, and that then—in […]Stephen Wolfram
  • 06
    Making Wolfram Tech Available as a Foundation Tool for LLM Systems
    Foundation Models Need a Foundation Tool LLMs don’t—and can’t—do everything. What they do is very impressive—and useful. It’s broad. And in many ways it’s human-like. But it’s not precise. And in the end it’s not about deep computation. So how can we supplement LLM foundation models? We need a foundation tool: a tool that’s broad […]Stephen Wolfram
  • 07
    What Ultimately Is There? Metaphysics and the Ruliad
    The Wolfram Institute recently received a grant from the Templeton World Charity Foundation for “Computational Metaphysics”. I wrote this piece in part as a launching point for discussions with experts in traditional philosophy. Moving Metaphysics from Philosophy to Science “What ultimately is there?” has always been seen as a fundamental—if thorny—question for philosophy, or perhaps […]Stephen Wolfram
  • 08
    P vs. NP and the Difficulty of Computation: A Ruliological Approach
    Empirical Theoretical Computer Science “Could there be a faster program for that?” It’s a fundamental type of question in theoretical computer science. But except in special cases, such a question has proved fiendishly difficult to answer. And, for example, in half a century, almost no progress has been made even on the rather coarse (though […]Stephen Wolfram
  • 09
    What Is Ruliology?
    Ruliology is taking off! And more and more people are talking about it. But what is ruliology? Since I invented the term, I decided I should write something to explain it. But then I realized: I actually already wrote something back in 2021 when I first invented the term. What I wrote back then was […]Stephen Wolfram
  • 10
    Instant Supercompute: Launching Wolfram Compute Services
    To immediately enable Wolfram Compute Services in Version 14.3 Wolfram Desktop systems, run RemoteBatchSubmissionEnvironment["WolframBatch"]. (The functionality is automatically available in the Wolfram Cloud.) Scaling Up Your Computations Let’s say you’ve done a computation in Wolfram Language. And now you want to scale it up. Maybe 1000x or more. Well, today we’ve released an extremely streamlined […]Stephen Wolfram
  • 11
    What’s Special about Life? Bulk Orchestration and the Rulial Ensemble in Biology and Beyond
    Towards a Theory of Bulk Orchestration It’s a key feature of living systems, perhaps even in some ways the key feature: that even right down to a molecular scale, things are orchestrated. Molecules (or at least large ones) don’t just move around randomly, like in a liquid or a gel. Instead, what molecular biology has […]Stephen Wolfram
  • 12
    The Ruliology of Lambdas
    Click any diagram to get Wolfram Language code to reproduce it. What Are Lambdas? It’s a story of pure, abstract computation. In fact, historically, one of the very first. But even though it’s something I for one have used in practice for nearly half a century, it’s not something that in all my years of […]Stephen Wolfram
共 12 条热搜查看完整榜单
  • 01
    Trading a Cloud Identity for Your Own: Workload Attestation on Managed Compute
    By Dhruv Pratap Introduction Organizations that have been around for a while usually run two identity systems side by side. One belongs to the cloud provider: IAM roles, instance profiles, execution roles. The other is your own, and it is the one your internal services actually check when they decide whether to answer a request. On infrastructure you build yourself, you can bootstrap your own identity however you like. On managed compute you cannot. The provider hands your process a cloud identiNetflix Technology Blog
  • 02
    Netflix Stratum Media Processing: Automated Container Right-Sizing
    By Violetta Pidvolotska and Naveen Mareddy Stratum is Netflix’s internal serverless media processing platform, shaped by over a decade of experience developing large-scale distributed systems. It powers a wide range of media workloads — including encoding, packaging, inspection, and quality scoring — behind playback, advertising, studio workflows, AI embeddings, and more. Engineers develop Stratum functions by packaging application code and OS dependencies into OCI images. Stratum then schedulesNetflix Technology Blog
  • 03
    Leave the Class Path in the Rearview Mirror
    Introducing composable, module system native and agent friendly command line tools for modern Java development By Danny Thomas, JVM Ecosystem Team Recent work on the Java language to pave the on-ramp has made it easier than ever to start a Java program and evolve it using the full language and platform. At the end of that on-ramp lies Java’s mature build and dependency management ecosystem, capable of carrying software to enormous scale and complexity. That ecosystem reached its maturity by deveNetflix Technology Blog
  • 04
    The Lifecycle of LLM-as-a-Judge: Building, Aligning, and Monitoring at scale
    By Emma Yanyang Kong , JJ Tan , Ishan Gupta , Lars Olds , Claire Campbell , David Fagnan , Ratna Kavuri , Rohan Gosain , Veli Balin , Minsu Jang , Louis Garcia Introduction Any team generating text with an LLM at scale hits the same problem. You cannot evaluate everything manually. You don't know which outputs are great, which are mediocre, and which are harmful, because you simply cannot read them all. The standard solution is LLM-as-a-Judge , a second model that scores the first model’s outputNetflix Technology Blog
  • 05
    Running Apache Spark experiments in my sleep (and on a plane)
    By Prashanth Gedde Narayanaswamy As one of Netflix’s largest and critical data pipelines, member-sessionizer consolidates tens of billions of hourly client events, encompassing user taps, playbacks, and engagement signals, into individual viewing sessions per profile. A quick two-week tuning effort spiraled into two months when standard approaches hit a wall. Because the failures only happened at full production scale, each test took hours, limiting me to just three or four attempts daily. The tNetflix Technology Blog
  • 06
    MAPS: Netflix’s Multimodal Asset Personalization at Scale
    By Emma Yanyang Kong , Aditya Deshpande , Asad Abbasi , Bowei Yan , David Fagnan , Ashish Rastogi , Dhaval Patel , Ray Zhang Introduction The Netflix experience is a journey of discovery. Every visual cue, from the artwork on a title to the video previews that autoplay while you browse, is there to connect you with a story you will love. We call these visual cues assets , and choosing the right one for each member is a personalization problem of its own. But which image or video preview of SquidNetflix Technology Blog
  • 07
    A Tale of Two Flink Autoscalers
    Samuel Yeboah , Francesco Di Chiara and Mingliang Liu Today, Netflix runs two Flink autoscalers. That is exactly one more than we want. We built the first one in-house years ago, when there was no mature option suited to our platform. The second came from the Apache Flink community, and it can scale workloads our homegrown system was never designed for. We now run both in production and are steadily converging on the open-source one. Along the way we learned some hard lessons about metrics, costNetflix Technology Blog
  • 08
    Netflix Conductor: The Next Chapter
    Netflix Conductor : The Next Chapter by Aravindan Ramkumar on behalf of the Conductor team Conductor is the workflow orchestration engine Netflix uses to stitch micro-services into reliable, observable business processes. If you’ve followed Conductor from the outside, the last thing you heard was probably the note on its GitHub repository : Netflix discontinued maintenance of Conductor OSS to refocus its resources on the internal fork. That announcement marked the end of the open-source chapter,Netflix Technology Blog
  • 09
    Behind the Scenes: Evolving Netflix’s Ads Event Pipeline for Live — Part II
    By Yogesh Nagarur Introduction In Part 1 of this series , we shared how Netflix built the Ads Event pipeline and approached it as a system design problem: replacing fragmented pipelines with a centralized collection, enrichment, and a standard data contract. The result was Ads Event Publisher, the system that collects ad telemetry from client devices, enriches it with the context of the ad that was actually served, and publishes a single, unified stream of ad events to every downstream consumer:Netflix Technology Blog
  • 10
    How and Why Netflix Built a Real-Time Distributed Graph: Part 3 — Querying the graph with gRPC…
    How and Why Netflix Built a Real-Time Distributed Graph: Part 3 — Querying the graph with gRPC execution API Authors: Nilesh Mishra and Ajit Koti This is the third entry of a multi-part blog series describing how we built a Real-Time Distributed Graph (RDG). In Part 1 , we discussed the motivation for creating the RDG and the architecture of the data processing pipeline that populates it. In Part 2 , we discussed how we designed the storage layer to handle billions of nodes and edges while maintNetflix Technology Blog
共 10 条热搜查看完整榜单
  • 01
    The Week in AI Careers: The Bottleneck Is People, Not Models, + More
    Anthropic put $100 million behind a diagnosis this week: enterprise AI adoption is bottlenecked by people who can deploy it , not by model capability. Draup data backed that reading from two directions, showing AI-specific roles climbing to 27 percent of Fortune 500 tech hiring and agent-orchestration postings surging on Wall Street. Engineering leaders at Shopify, Cockroach Labs, and Airbnb each pointed to a version of the same shift in their own operations this week. Generalist engineers are cODSC - Open Data Science
  • 02
    AI Infrastructure, Agent Safety, and Developer Skills
    The ODSC AI West Full Schedule is live! With 150+ hands-on sessions across 9 tracks, three days fill up quickly. See what’s happening across agentic AI, AI engineering, evals and reliability, data infrastructure, Physical AI, machine learning, developer tools, AI leadership, and more. Register now with 40% off AI-Assisted Coding Is Changing What It Means to Be a Good Developer AI-assisted coding can generate code quickly, but strong developers still define problems, test results, and own what shODSC - Open Data Science
  • 03
    From AI User to AI Builder: Start with Skills
    Most of us have become reasonably good at using AI. We prompt, we refine, we upload documents. Increasingly we connect AI to search and to the tools we already work in: Gmail, Slack, Drive, Notion, GitHub, the Salesforce CRM. The results can be genuinely good and useful. But there is still a gap between using AI well and building with AI. The difference is: an AI user gets a good result. an AI builder creates a process that produces that good result again and again, and without anyone sitting inODSC - Open Data Science
  • 04
    Week in AI Careers: AI Skills Become the Third-Hardest to Recruit For
    The dominant careers signal last week was a widening mismatch between how AI-ready workers believe they are and what employers actually expect. A new survey from Express Employment Professionals and The Harris Poll found that nearly all job seekers consider themselves AI-ready while employers keep raising the bar toward advanced skills, and a Payscale preview reported that pay structures are lagging the roles companies are rewriting around AI. European hiring data from The Stepstone Group showedODSC - Open Data Science
  • 05
    AI-Assisted Coding Is Changing What It Means to Be a Good Developer
    What makes someone a good developer when an AI tool can write a working function in seconds? The answer has less to do with how quickly someone types and more to do with whether they can turn a real problem into reliable software. AI-assisted coding changes how developers produce code. It also puts more weight on the skills that surround it: defining requirements, understanding architecture, testing assumptions, reviewing changes, and deciding what should ship. A developer can now ask an AI toolODSC - Open Data Science
  • 06
    AI Incidents, Agent Memory, and AI Careers
    AI for Work Summit NYC Learn how to apply AI in real work with two days of practical sessions, workshops, and expert instruction in NYC. Register this week to secure Super Early Bird pricing before rates go up. Learn directly from experts like Dr. Jon Krohn, Paige Bailey, Michael Schrage, Christine Long, Sid Dhulipalla, and Michael Levan . Save up to $600 Trump-Xi Summit Could Open AI Safety Talks as Competition Continues Trump and Xi may consider an AI incident notification channel as cyber risODSC - Open Data Science
  • 07
    The Week in AI Careers: Workers Expect Job Losses, Executives Say Otherwise, + More
    The clearest careers signal from the week came from Pew Research Center, which found that across most of the world people now expect AI to shrink the number of jobs rather than grow it. A companion Pew release showed the worry has flipped along party lines in the United States, with Democrats now more concerned about AI and jobs than Republicans, a reversal from two years ago. Running against that public mood, executives on stage at Salesforce’s Dreamforce conference spent the week arguing the oODSC - Open Data Science
  • 08
    How AI Startups Can Get More ROI From AI Conferences
    What makes an AI conference worth the cost for a startup? The answer is not the number of badge scans, business cards, or booth visitors collected over three days. Strong AI conference ROI comes from converting event access into faster technical decisions, qualified partnerships, recruiting signals, and commercial opportunities that continue moving after the venue closes. Why AI Conference ROI Requires a Different Model Conference participation can become expensive quickly. Registration or sponsODSC - Open Data Science
  • 09
    Apple’s Siri AI, Anthropic’s Safety Move, and This Week’s Top AI Stories
    AI for Work Summit NYC Join us for a two-day, workshop-driven event dedicated to making AI practical at work. Choose from four tracks ( Build, Apply, Lead, and AI Engineering Fundamentals ) and learn directly from experts like Dr. Jon Krohn, Paige Bailey, Michael Schrage, Christine Long, Sid Dhulipalla, and Michael Levan . Register with Super Early Bird Price Anthropic Disrupts Claude Use That Could Support Biological Weapons Development What did Anthropic discover about the potential misuse ofODSC - Open Data Science
  • 10
    What Microsoft’s Openclaw Reversal Means for Picking a Harness
    In March 2026, Satya Nadella told a Morgan Stanley audience he admired OpenClaw as a piece of engineering, then added that shipping it inside Microsoft would amount to launching a virus . By the first week of June, its creator, Peter Steinberger, was on stage with Nadella at Microsoft Build , helping unveil a Windows version of the same tool, sandboxed and rebranded as Scout. Brad Groux opened his ODSC talk with that reversal, not to talk about Microsoft, but to make a point about how fast the gODSC - Open Data Science
共 10 条热搜查看完整榜单
  • 01
    Computational Chemistry with Wolfram: From Learning to Research
    Modern chemistry depends on computation at nearly every stage of the scientific process. Chemists begin by representing molecular structures, then analyze chemical data, investigate molecular behavior and visualize the results of their work. Those tasks require computational tools that support scientific workflows rather than isolated calculations. Today, computation is as fundamental to chemistry as the […]Mark Long
  • 02
    Launching Version 15 of Wolfram Language & Mathematica: Built-in (Useful) AI & Lots of New Core Functionality
    June 23, 1988 is when we launched Version 1.0 of Mathematica. Today—almost 38 years later—we’re launching Version 15 of what—in recognition of how far it’s expanded beyond “math”—we now call Wolfram Language. It’s an impressive release, with a lot of new core functionality. It might perhaps seem surprising that after 38 years there’d still be […]Stephen Wolfram
  • 03
    OpenAI Disproves Erdős Unit Distance Conjecture
    The Erdős unit distance problem asks for the largest possible number u ( n ) of unit distances among n points in the plane. This is equivalent to finding maximally dense unit-distance graphs. A recent OpenAI announcement concerns the asymptotic problem: the old n ^(1+ o (1)) expectation is false.Wolfram Blog Team
  • 04
    Theory Meets Practice: 8 Books to Start Your Career in Astrophysics, Geography, Civil Engineering and More
    In or out of school, the opportunities to learn and grow in your career are endless, and Wolfram is proud to bolster those with educational resources, from courses to textbooks. We are happy to share conversations with two authors whose books cover applications of Wolfram technology in astrophysics and geography, as well as highlight a […]Treyton Jansen
  • 05
    Compression and Recompression of JPEG: Stability, Artifacts and Iterative Image Collapse
    Every semester or two, students ask me about the implications of re-compression, especially of JPEG files. Compression comes in two main kinds—Lossless, which is completely invertible, meaning you get exactly what you put in when you decompress, and Lossy which ‘throws out’ things deemed unimportant in some way, so that when you decompress, you get […]Wolfram Blog Team
  • 06
    A Data Adventure in Boston, 1929: Historical Census Corpus Analysis
    I am writing a novel. It’s a historical fiction thing. Apparently, that means I need to do a lot of research on what life was like in the 1920s. ​ My problem last night was, my character moves to Boston from Chicago, and in order to give the city texture, we need to introduce characters, […]Wolfram Blog Team
  • 07
    LLMs, Symbolic Computation and the Future of Mathematical Discovery
    “The cat’s out of the bag,” said the mathematician Andrew Granville, reflecting on the rapid improvement of AI systems. His phrase captures the mood of the moment: by 2025-26, large language models (LLMs) had become powerful enough to move from impressive demonstrations to serious mathematical and scientific use. AI systems reached gold-medal level at the […]Wolfram Blog Team
  • 08
    Computational Geometry Modeling of the Neolithic Circular Ditch in Vinoř, Prague
    This tutorial is a follow-up to a recent post by the author herself about archeoastronomical modeling of Central European Neolithic Circular Ditches [1], or roundels, with Wolfram 3D graphical primitives. Here, the focus will be instead on the use of mesh-based primitives from computational geometry to build a realistic 3D model of a roundel recently […]Wolfram Blog Team
  • 09
    A Modern eTextbook on Laplace Transforms for Engineering, Science and More
    The Laplace transform is such an effective tool for solving problems in the fields of science and engineering—it’s one of the main tools available for solving both ordinary differential equations (ODEs) and partial differential equations (PDEs). I’m excited to announce that the notebook version of Laplace Transforms in Theory and Practice: A Computational Approach by Hrachya Khachatryan […]Juan Ortiz
共 9 条热搜查看完整榜单
  • 01
    Issue 769
    iPhone Duo Group Lab, McKinley, SwiftFairy, ArrangementViewnewsletter@iosdevweekly.com (next app's swiftCon team)
  • 02
    Issue 767
    Homework for a folding iPhone, Shopify goes back to native, and a login keychain you can't back upnewsletter@iosdevweekly.com (next app's swiftCon team)
  • 03
    Issue 766
    John Ternus becomes Apple's new CEO, App Store leadership changes, boosting Simulator performance with SimSlim, and nested weak self capture warnings in Swift.newsletter@iosdevweekly.com (next app's swiftCon team)
  • 04
    Issue 765
    UISceneDelegate mandate, M6 Mac mini, Apple layoffs, Headless Xcode MCP, Embedded Swift 6.4, SwiftTUInewsletter@iosdevweekly.com (next app's swiftCon team)
  • 05
    Issue 764
    AI judgement, observability, SwiftPM traits and package registriesnewsletter@iosdevweekly.com (next app's swiftCon team)
  • 06
    Issue 762
    OpenAI v Apple, Databases, ContentBuildernewsletter@iosdevweekly.com (next app's swiftCon team)
  • 07
    Issue 761
    Apple Upgrade, Modularization, AI Code Review, Approachable Concurrency, Vision Pronewsletter@iosdevweekly.com (next app's swiftCon team)
  • 08
    Issue 760
    Claude Code gets an iOS Simulator, SwiftData sync gets more practical, and Xcode 27 keeps getting better.newsletter@iosdevweekly.com (next app's swiftCon team)
  • 09
    Issue 759
    Apple sues OpenAI, SwiftUI environment footguns, and Xcode’s MCP tools without an agentnewsletter@iosdevweekly.com (next app's swiftCon team)
  • 10
    Issue 758
    Xcode 27 Beta 3, OpenAI Sol release, Fable 5 extensionsnewsletter@iosdevweekly.com (next app's swiftCon team)
  • 11
    Issue 757
    Swift 6.3.3 release, Apple hardware price rises, Xcode 27 compatibility fixes, SwiftUI internals, design systems, agent-friendly logging, and Swift on an Apple II.newsletter@iosdevweekly.com (next app's swiftCon team)
  • 12
    Issue 756
    Swift Package Index joins Apple; Dave Verwer joins Apple toonewsletter@iosdevweekly.com (next app's swiftCon team)
  • 13
    Issue 755
    Slicing iOS launch times, custom Xcode agent skills, and a nostalgic look back at vintage Intel Macs. 🖥️newsletter@iosdevweekly.com (next app's swiftCon team)
  • 14
    Issue 754
    Before every WWDC, developers don't just make predictions, they play an unofficial game of WWDC Bingo. Read on to see if any of your squares got checked off.newsletter@iosdevweekly.com (next app's swiftCon team)
  • 15
    Issue 753
    WWDC 2026 week: Apple's AI credibility test, developer wishlists, SwiftUI animation timing, Core Data + Observation, and Swift concurrency deep dives.newsletter@iosdevweekly.com (next app's swiftCon team)
  • 16
    Issue 752
    WWDC26 anticipation builds alongside deep dives into SwiftData, custom layouts, and open-source AI agent tooling.newsletter@iosdevweekly.com (next app's swiftCon team)
  • 17
    Issue 751
    End of an era and new beginnings. Plus understanding animations, feature flags, and hot reloads.newsletter@iosdevweekly.com (next app's swiftCon team)
  • 18
    Issue 750
    From tinkering with Objective-C to writing a little newsletter for almost fifteen years. 😱newsletter@iosdevweekly.com (next app's swiftCon team)
  • 19
    Issue 749
    There’s a new community-driven index for Swift playgrounds, and it’s great.newsletter@iosdevweekly.com (next app's swiftCon team)
  • 20
    Issue 748
    What’s your position on LLM-generated pull requests in open source projects? 🤖newsletter@iosdevweekly.com (next app's swiftCon team)
  • 21
    Issue 747
    Is Swift growing bigger than Xcode? Yes, and it has been for a while! 🚀newsletter@iosdevweekly.com (next app's swiftCon team)
  • 22
    Issue 746
    Dubbbb Dubbbbbbbbbbbb! 🚀newsletter@iosdevweekly.com (next app's swiftCon team)
  • 23
    Issue 745
    How can you get your new blog started? If only there were some kind of directory to list it in. 🎯newsletter@iosdevweekly.com (next app's swiftCon team)
  • 24
    Issue 744
    How do we avoid software being average if we let AI write everything? 🎢newsletter@iosdevweekly.com (next app's swiftCon team)
  • 25
    Issue 743
    Should an agent and your instructions to it live in the sidebar, or should it be front and center? 🤖newsletter@iosdevweekly.com (next app's swiftCon team)
  • 26
    Issue 742
    Did I find a new flagship project for Swift on Windows? Maybe! 🫨newsletter@iosdevweekly.com (next app's swiftCon team)
  • 27
    Issue 741
    Forget about Windows and Wasm, what’s the flagship project for macOS and iOS?! 😂newsletter@iosdevweekly.com (next app's swiftCon team)
  • 28
    Issue 740
    Want to use an agent, but without launching a terminal or an IDE? 🫨newsletter@iosdevweekly.com (next app's swiftCon team)
  • 29
    Issue 739
    Cupertino? Isn’t that a city in California? 🐻newsletter@iosdevweekly.com (next app's swiftCon team)
  • 30
    Issue 738
    Alternate browser engines come to the Japanese App Store, but what will it take to see this in use? 🤔newsletter@iosdevweekly.com (next app's swiftCon team)
  • 31
    Issue 737
    Last issue of the year, and it’s time for the App Store Awards. 🎉newsletter@iosdevweekly.com (next app's swiftCon team)
  • 32
    Issue 736
    Can LLMs generate not only content and data, but user interfaces, too? 🤔newsletter@iosdevweekly.com (next app's swiftCon team)
  • 33
    Issue 735
    Can we briefly talk about keeping track of the Swift open-source project? 💼newsletter@iosdevweekly.com (next app's swiftCon team)
  • 34
    Issue 734
    Did you have a web-based view of the App Store storefront on *your* bingo card?newsletter@iosdevweekly.com (next app's swiftCon team)
  • 35
    Issue 733
    Swift on Android is now included in nightly Swift builds, will it be official in 6.3?newsletter@iosdevweekly.com (next app's swiftCon team)
  • 36
    Issue 732
    Why did Liquid Glass start therapy? To work through its transparency issues. 😅newsletter@iosdevweekly.com (next app's swiftCon team)
  • 37
    Issue 731
    What’s the verdict on Liquid Glass now it’s out in the wild? 🧊newsletter@iosdevweekly.com (next app's swiftCon team)
  • 38
    Issue 730
    Will you forgive me one more issue where I talk about Swift on the server? 😬newsletter@iosdevweekly.com (next app's swiftCon team)
  • 39
    Issue 729
    Where did the Swift and Apple platform community go? 🧐newsletter@iosdevweekly.com (next app's swiftCon team)
  • 40
    Issue 728
    Ten years of Swift on the server is a milestone worth celebrating, but where do we go from here? 🤔newsletter@iosdevweekly.com (next app's swiftCon team)
  • 41
    Issue 727
    Should a design case study hosted by Apple be part of the Apple Design Award prize? 🤔newsletter@iosdevweekly.com (next app's swiftCon team)
  • 42
    Issue 726
    It’s not as flashy as a vapour chamber, but it’ll keep you safer! 🦺newsletter@iosdevweekly.com (next app's swiftCon team)
  • 43
    Issue 725
    Are you all ready for Tuesday? It’s Liquid Glass time! 🧊newsletter@iosdevweekly.com (next app's swiftCon team)
  • 44
    Issue 724
    Have we seen the last changes to Xcode 26 before the public release? I think we have now!newsletter@iosdevweekly.com (next app's swiftCon team)
  • 45
    Issue 723
    What is Apple not saying about Liquid Glass? 🧊newsletter@iosdevweekly.com (next app's swiftCon team)
  • 46
    Issue 722
    What does Seb Vidal’s post tell us about the future of UIKit? 🤔newsletter@iosdevweekly.com (next app's swiftCon team)
  • 47
    Issue 721
    At what point does something become “written by an LLM”? 🤖newsletter@iosdevweekly.com (next app's swiftCon team)
  • 48
    Issue 720
    We’ll see accessibility “nutrition labels” on App Store listings in a few months, but will we see more in the future?newsletter@iosdevweekly.com (next app's swiftCon team)
  • 49
    Issue 719
    What is the Foundation Model good at, and how does it compare to the alternatives?newsletter@iosdevweekly.com (next app's swiftCon team)
  • 50
    Issue 718
    Are you using the Foundation Model APIs yet? Let me know what you’re doing with them! 🤖newsletter@iosdevweekly.com (next app's swiftCon team)
共 50 条热搜查看完整榜单
  • 01
    HELM Arabic Enterprise
    We present HELM Arabic Enterprise, a leaderboard for transparent, reproducible evaluation of large language models on Arabic-language benchmarks designed around enterprise use cases. The leaderboard was developed in collaboration with Arabic.AI and builds on the HELM evaluation methodology: standardized prompting, fully logged requests and responses, and reproducible scoring through the open-source HELM framework.Yifan Mai
  • 02
    HELM Arabic
    As part of our efforts to better understand the multilingual capabilities of large language models (LLMs), we present HELM Arabic, a leaderboard for transparent and reproducible evaluation of LLMs on Arabic language benchmarks. This leaderboard was produced in collaboration with Arabic.AI.Yifan Mai
  • 03
    HELM Long Context
    We introduce the HELM Long Context leaderboard for transparent, comparable and reproducible evaluations of long context capabilities of recent models. Introduction Recent Large Language Models (LLMs) support processing long inputs with hundreds of thousands or millions of tokens. Long context capabilities are important for many real-world applications, such as processing long text documents, conducting long conversations or following complex instructions. However, support for long inputs does noYifan Mai
  • 04
    Reliable and Efficient Amortized Model-Based Evaluation
    TLDR: We enhance the reliability and efficiency of language model evaluation by introducing IRT-based adaptive testing, which has been integrated into the HELM framework.Sang Truong
  • 05
    Surprisingly Fast AI-Generated Kernels We Didn’t Mean to Publish (Yet)
    TL;DRAnne Ouyang
  • 06
    BountyBench: Dollar Impact of AI Agent Attackers and Defenders on Real-World Cybersecurity Systems
    We introduce BountyBench , a benchmark featuring 25 systems with complex, real-world codebases, and 40 bug bounties that cover 9 of the OWASP Top 10 Risks. Key Takeaways BountyBench is a benchmark containing 25 diverse systems and 40 bug bounties, with monetary awards ranging from $10 to $30,485, covering 9 of the OWASP Top 10 Risks . It is designed to evaluate offensive and defensive cyber-capabilities in evolving real-world systems. To capture the vulnerability lifecycle from discovery to repaAndy K. Zhang
  • 07
    HELM Capabilities: Evaluating LMs Capability by Capability
    Introducing HELM Capabilities, a benchmark that evaluates language models across a curated set of key capabilities, providing a comparison of their strengths and weaknesses. Evaluating language models is a dynamic and critical process as models continue to improve rapidly. Understanding their strengths and weaknesses is essential for external users to determine which models suits their needs. Two years ago, we introduced the Holistic Evaluation of Language Models (HELM) as a framework to assessJialiang Xu
  • 08
    General-Purpose AI Needs Coordinated Flaw Reporting
    Today, we are calling for AI developers to invest in the needs of third-party, independent researchers, who investigate flaws in AI systems. Our new paper advocates for a new standard of researcher protections, reporting and coordination infrastructure. The paper, In House Evaluation Is Not Enough: Towards Robust Third-Party Flaw Disclosure for General-Purpose AI, has 34 authors with expertise in machine learning, law, security, social science, and policy. Introduction Today, we are calling forShayne Longpre
  • 09
    HELM Safety: Towards Standardized Safety Evaluations of Language Models
    *Work done while at Stanford CRFMFarzaan Kaiyom
  • 10
    Advancing Customizable Benchmarking in HELM via Unitxt Integration
    The Holistic Evaluation of Language Models (HELM) framework is an open source framework for reproducible and transparent benchmarking of language models that is widely adopted by academia and industry. To meet HELM users’ needs for more powerful benchmarking features, we are proud to announce our collaboration with Unitxt, an open-source community platform developed by IBM Research for data preprocessing and benchmark customization. The integration of Unitxt into HELM gives HELM users access toYifan Mai
共 10 条热搜查看完整榜单