[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-4761":3,"related-4761":71},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":9,"source_name":10,"source_url":8,"published_at":11,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":15,"score_detail":16,"sources":24,"tags":26,"primary_topic":8,"secondary_topics":32,"classification":33,"search_phrases":65,"slug":68,"view_count":69,"doi":8,"paper":8,"created_at":70},4761,"\"AI+农业\":如何从\"云\"上落到\"地\"里","https:\u002F\u002Fwww.agri.cn\u002Fzt\u002Fzhzz\u002Fzhsc\u002F202610\u002Ft20261010_8875651.htm","农民日报记者走访内蒙古、河北、江苏等地一线调研。展示蒙牛嘉立荷牧场牛脸识别、河北卢龙78座高标准智能温室、京津冀植保无人机、潍坊科技学院自动栽苗机器人等鲜活案例,呈现智慧农业从会种会养走向慧种慧养的产业图景。据智研咨询,2025年我国智慧农业市场规模约1200亿元,年复合增长率15.28%。文章呼吁打通技术成本与农户算账问题。",null,"内蒙古和林格尔县的蒙牛嘉立荷生态牧场，和江苏连云港的东旺奶牛养殖基地，正运行着同一套“牛脸识别”系统。无需近距离拍摄，后台屏幕动态框出牛脸，饮食、健康、行为数据一目了然。\n\n“这不是保险理赔用的识别工具，而是智慧养殖工具。”系统研发方寿光市翔天农业科技有限公司副总经理刘守城说。每头牛都有“数字身份证”，全天候“点名”，可实现精确投喂。养殖数据直传农业、防疫部门，食品安全可追溯。试点以来，已覆盖超7000头牛，饲料利用率提升25%，发病率降低20%。","农民日报 中国农业农村信息网 2026-10-10","2026-10-10T00:00:00Z","报道",10,false,72,{"impact":17,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":23},18,20,15,12,7,1,"以牛脸识别落地案例呈现AI养殖的实效数据，兼具产业参考价值与可复制经验。",[25],{"name":10,"url":6},[27,28,29,30,31],"智慧农业","农业人工智能","智慧养殖","食品安全追溯","牛脸识别",[],{"status":34,"locked":14,"version":35,"method":36,"evidence":37},"needs_review","2026-10-22.2","rules",{"candidates":38,"agricultural_evidence":42,"reason":64},[39,46,53,59],{"code":40,"score":41,"supported":42,"title":43,"abstract":44,"tags":45},"I1",4,true,[],[27],[27],{"code":47,"score":48,"supported":42,"title":49,"abstract":50,"tags":52},"B7",3,[],[51],"温室",[],{"code":54,"score":48,"supported":42,"title":55,"abstract":56,"tags":58},"F4",[],[57],"植保",[],{"code":60,"score":22,"supported":14,"title":61,"abstract":62,"tags":63},"I2",[],[],[28],"多个产业证据相当，需要确认主归属",[66,67],"蒙牛 嘉立荷 牛脸识别","翔天农业 智慧养殖","蒙牛嘉立荷牛脸识别-4761",2,"2026-10-13T00:08:33.631486Z",{"total":72,"page":22,"page_size":72,"items":73},6,[74,131,187,230,294,359],{"id":75,"title":76,"url":77,"summary":78,"summary_zh":8,"content":79,"source_name":80,"source_url":8,"published_at":81,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":82,"score_detail":83,"sources":87,"tags":89,"primary_topic":92,"secondary_topics":98,"classification":99,"search_phrases":126,"slug":129,"view_count":21,"doi":8,"paper":8,"created_at":130},3387,"从会种粮到'慧'种粮——中联重科岳麓智慧农场、常州艾宝机器人、温氏华系种猪、牧原智能猪舍AI兽医集体亮相","https:\u002F\u002Fnews.cyol.com\u002Fgb\u002Fkeji\u002Farticles\u002F2026-09\u002F23\u002Fcontent_v6vKznCP2x.html","中联重科岳麓智慧农场'天空地人机'协同智慧作业模式搭载多光谱相机巡田无人机最高飞120米、20分钟完成整片地块巡检，长势图精度达3米×3米网格；常州艾宝机器人四臂采摘猕猴桃机器人经第三方检测适合山坡丘陵地形，一台机器能节约7至8个人工；温氏股份、德康农牧'华系种猪育种技术与核心种源创制及应用'项目获国家科学技术进步奖一等奖；牧原股份智能化猪舍AI兽医全天候值守。","◎记者 刘立 王乔琪\n\n又是一年丰收季。田野之间，沉甸甸的果实背后，农业生产逻辑正在发生深刻变化。\n\n今年5月，国务院印发的《加快农业农村现代化“十五五”规划》明确提出：“推进人工智能运用和智慧农业发展。加快农业人工智能技术创新、应用场景拓展和产业生态营造。”\n\n近期，上海证券报记者在对部分农业种养殖龙头企业和农业展会的调研中发现，从种子实验室到果园大田，智能化工具正在把农业从经验驱动推向数据驱动。AI不再只是实验室概念，它正从育种、田间种植延伸到畜牧水产全链条，改写着传统农业依靠经验、看天吃饭的旧模式。\n\n**以AI打磨农业“芯片”**\n\n种业被称作农业的“芯片”。记者在采访中发现，快速发展的AI技术，正在缩短育种探索的漫长周期。\n\n在近日举办的全国种子双交会上，中种集团展台上一台ELISA蛋白全自动分析检测工作站设备吸引不少参会者驻足。据工作人员介绍，传统育种蛋白检测多采用人工、半自动化操作，通量低、稳定性差，而这台设备针对生物育种性状鉴定、蛋白定量筛查场景，可实现实验全流程无人自动化运行，日检测通量比人工高5倍以上，成本降低50%，且完全国产化。\n\n记者在采访中发现，AI正在把复杂的农业科研数据转化成通俗直观的结论。育种不再完全依靠科研人员肉眼观察记录，而是由仪器采集数据，AI完成解析筛选。\n\n上海泽泉科技股份有限公司研发的花粉活力仪可以应用于玉米、小麦、果树、大豆杂交育种项目，帮助科研人员精准把握授粉窗口，稳定后代品质。公司副总经理张凯峰对记者说：“过去做大豆、玉米、小麦等杂交育种，收集花粉之后，不知道花粉活性。我们的设备可以对每一粒花粉活性做量化测定，活性不达标的花粉直接淘汰。随着科技的普及，我相信，未来这些智能设备都会逐步下沉到普通农户。”\n\n一位农业专家对记者表示，AI并非颠覆传统育种，而是对育种体系的升级赋能。智能育种的核心，是依靠高通量设备获取标准化的基因组、表型组数据，依托算法挖掘基因与性状之间的关联，推动育种从过去依赖经验的“试错筛选”，转向数据驱动的“精准设计”，助力我国种业向育种4.0阶段迈进。\n\n**“空天地一体化”重构农业生产模式**\n\n在位于湖南长沙的中联重科岳麓智慧农场，中联智慧农业打造的“天空地人机”协同智慧作业模式，正在改写大田管护方式。\n\n以往人工巡查四五百亩稻田，要耗费一至两小时。中联智慧农业这套搭载多光谱相机的巡田无人机最高可飞120米，20分钟就能完成整片地块巡检，变焦可达100多倍，捕捉到虫害、长势异常等疑点便低空补拍核验。影像实时回传云端后，AI算法输出水稻长势图，精度可达3米乘以3米网格，明确标注地块补水、补肥位置与用量。\n\n天上完成巡查，地上设备随即落地执行。中联智慧农业部署的物联网灌排设备、微型气象站同步采集风、雨、光照数据，农户通过手机软件就能查看农事日历、操控田间设备。这种“按数据决策”的模式，改变了传统农业“凭经验跑田”的粗放管理方式。\n\n除了天上飞的无人机，地上走的机器人也开始应用到更多农业生产场景。\n\n在北京平谷的2026世界农业科技创新大会上，常州艾宝机器人有限公司研发的猕猴桃采摘机器人吸引了不少观众。公司工程师赵炳宇向记者介绍，这款四臂采摘机器人，经第三方机构检测，适合山坡丘陵地形，一台机器能够节约7至8个人工。现阶段设备尚处在定制阶段，已有3家企业完成定制采购。\n\n“我们主要从事AI智能水果采摘机器人研发，已突破多臂协同、视觉感知、自主导航等关键技术，已经推出苹果、葡萄、猕猴桃采摘等系列产品，目的就是推动采摘机器人从技术示范走向生产落地。”赵炳宇说。\n\n**农牧业智能化驶入快车道**\n\n中国是全球最大的猪肉消费国，年消费猪肉超过5000万吨。庞大的产业规模背后，是管理规模大、场景复杂的生猪养殖体系。如今，人工智能正加速渗透进这个最传统的行业。\n\n畜牧种业是畜牧业的“芯片”，在很长一段时间，国内要从海外进口种猪，还要高价采购海外生猪基因芯片，支付高昂的知识产权费。通常一头外引种猪的综合成本高达3万元至5万元。\n\n这一局面如今已被打破。日前，温氏股份、德康农牧等作为主要完成单位参与的“华系种猪育种技术与核心种源创制及应用”项目，荣获国家科学技术进步奖一等奖。该项目通过基因组选择、表型精准测定等技术创新，构建了华系种猪自主育种体系，创制了具有自主知识产权的核心种猪群体，标志着我国生猪核心种源实现高标准自主可控，彻底打破国外长期垄断格局。\n\n在河南，畜牧业龙头企业牧原股份的智能化猪舍里，AI兽医已投入全天候值守。\n\n“过去发现猪不吃食，饲养员找兽医、翻手册、凭经验判断，平均响应时间需要个把小时；现在从发现异常到推荐用药的闭环，仅需数秒。”牧原股份技术负责人介绍，名为“小牧助手”的AI应用，融合企业30余年养殖知识与全产业链330万套智能装备每日产生的20亿条数据，被封装为可嵌入日常流程的AI智能体。目前，牧原与阿里云合作共建的养猪大模型“猪小牧”已向全行业开放共享，助力行业高质量发展。\n\n此外，类似变革正在更多养殖场景中铺开。比如在水产养殖领域，广西防城港东兴市红树林农业有限公司的养殖基地引入AI“厂长”，集成AI算法、高清水下摄像头与传感器网络，24小时实时采集虾群数据，自动调控水阀与投料，每日精准投喂300余次。目前，该基地一年可养对虾5茬，年产量35万至40万公斤。\n\n【责任编辑：母建鑫】","中青在线 2026-09-23","2026-09-23T00:00:00Z",85,{"impact":84,"substance":85,"depth":17,"authority":20,"freshness":21,"relevant":22,"comment":86},26,22,"紧扣十五五规划AI+农业主线，覆盖育种、种植、养殖全链条，案例与数据扎实，具行业风向标价值。",[88],{"name":80,"url":77},[27,28,90,91,29],"智能农机","种业振兴",{"code":93,"slug":94,"name":95,"path":96,"parent_code":97},"B2","fruit","果树","\u002Ftopic\u002Fhorticulture\u002Ffruit","B",[],{"status":100,"locked":14,"version":35,"method":36,"evidence":101},"classified",{"candidates":102,"agricultural_evidence":42},[103,108,113,118,122],{"code":93,"score":48,"supported":42,"title":104,"abstract":105,"tags":107},[],[106],"猕猴桃",[],{"code":109,"score":22,"supported":14,"title":110,"abstract":111,"tags":112},"F1",[],[],[91],{"code":114,"score":22,"supported":14,"title":115,"abstract":116,"tags":117},"F5",[],[],[90],{"code":40,"score":22,"supported":14,"title":119,"abstract":120,"tags":121},[],[],[27],{"code":60,"score":22,"supported":14,"title":123,"abstract":124,"tags":125},[],[],[28],[127,128],"中联重科 岳麓智慧农场","温氏 华系种猪","中联重科岳麓智慧农场-3387","2026-09-25T00:09:27.645047Z",{"id":132,"title":133,"url":134,"summary":135,"summary_zh":8,"content":8,"source_name":136,"source_url":8,"published_at":137,"category":138,"cover_url":8,"hotness":13,"is_selected":14,"score":139,"score_detail":140,"sources":145,"tags":147,"primary_topic":150,"secondary_topics":156,"classification":157,"search_phrases":172,"slug":175,"view_count":41,"doi":8,"paper":176,"created_at":186},3322,"PB-DETR：群养生猪多行为识别的自适应液体神经网络编码器与动态监督框架","https:\u002F\u002Fjky.yangtzeu.edu.cn\u002Finfo\u002F1265\u002F6884.htm","长江大学计算机科学学院研究生郭宇衡作为第一作者，在计算机与农业交叉领域顶级期刊《Computers and Electronics in Agriculture》发表题为PB-DETR: multi-behaviour recognition of group-housed pigs with adaptive liquid neural network encoder and dynamic supervision的研究论文。长江大学智慧农业交叉创新研究中心主任、计科学院詹炜教授为通讯作者。研究将液体神经网络机制引入Transformer编码器，设计自适应液体时间常数增强模块，构建多尺度特征交互和动态质量感知优化策略，进而降低模型对固定背景信息的依赖。基于PB-DETR框架构建的模型平均精度(mAP)达到0.893，参数量为14.58M，计算量为49.12GFLOPs；在不同养殖场景之间的测试中表现出较好稳定性。","《Computers and Electronics in Agriculture》2026","2026-09-21T00:00:00Z","论文",78,{"impact":141,"substance":85,"depth":17,"authority":142,"freshness":143,"relevant":22,"comment":144},16,14,8,"农业人工智能顶刊论文，方法新颖、指标明确，对智慧养殖行为识别有实质参考价值，值得进入每日精选。",[146],{"name":136,"url":134},[27,28,29,148,149],"目标检测","生猪行为识别",{"code":151,"slug":152,"name":153,"path":154,"parent_code":155},"C1","pigs","生猪","\u002Ftopic\u002Flivestock\u002Fpigs","C",[],{"status":100,"locked":14,"version":35,"method":36,"evidence":158},{"candidates":159,"agricultural_evidence":42},[160,164,168],{"code":151,"score":143,"supported":42,"title":161,"abstract":162,"tags":163},[153],[],[],{"code":40,"score":41,"supported":42,"title":165,"abstract":166,"tags":167},[],[27],[27],{"code":60,"score":69,"supported":14,"title":169,"abstract":170,"tags":171},[],[],[28,148],[173,174],"长江大学 詹炜 生猪行为识别","PB-DETR 群养生猪 多行为识别","长江大学詹炜生猪行为识别-3322",{"doi":8,"openalex_id":8,"authors":177,"venue":8,"cited_by_count":178,"oa_url":8,"card":179,"direction":183,"ingested_from":185},[],0,{"tldr":180,"method":181,"finding":182,"direction":183,"opportunity":184},"提出PB-DETR框架，实现群养生猪多行为识别，mAP达0.893。","引入液体神经网络改进Transformer编码器，结合多尺度特征交互与动态监督。","模型在群养生猪多行为识别中mAP达0.893，跨场景稳定性较好。","农业人工智能与决策模型","可探索液体神经网络在其它畜禽行为识别中的泛化能力及轻量化部署。","agent","2026-09-24T00:04:02.413623Z",{"id":188,"title":189,"url":190,"summary":191,"summary_zh":8,"content":192,"source_name":193,"source_url":8,"published_at":194,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":195,"score_detail":196,"sources":198,"tags":200,"primary_topic":203,"secondary_topics":208,"classification":209,"search_phrases":225,"slug":228,"view_count":213,"doi":8,"paper":8,"created_at":229},126,"AI+农业如何从云上落到地里——从\"会养\"\"会种\"到\"慧养\"\"慧种\",2025年农业人工智能Token年度消耗超2.4亿","https:\u002F\u002Ffinance.sina.com.cn\u002Fjjxw\u002F2026-07-27\u002Fdoc-inikekzt3337234.shtml","新华社报道:人工智能正深度融入农田,从温室喷雾机器人、牛脸识别、北大荒多光谱无人机AI查苗到河北卢龙78座高标准智能温室蓝莓基地,各地探索\"慧养\"\"慧种\"。2025年中国智慧农业市场规模从2017年384.74亿元增至1200亿元,年复合增长率15.28%。但产业仍面临三重短板:一是初始投入大(智慧大棚每平方米300-1500元,种植集装箱20-30万元);二是农村主力为中老年群体,数字化操作能力不足、青年人才短缺;三是山区丘陵复杂地形数据与算法迁移困难,许多AI应用仍停留在监测识别预警阶段。寿光\"AI机器人银发专家\"将40年寿光模式经验提炼为数智化内核,技术服务效率提升5倍以上,业内认为\"让技术找人\"比\"让人学技术\"更易实现。","内蒙古自治区和林格尔县的蒙牛嘉立荷生态牧场，和江苏连云港的东旺奶牛养殖基地，正运行着同一套“牛脸识别”系统。无需近距离拍摄，后台屏幕动态框出牛脸，饮食、健康、行为数据一目了然。\n\n“这不是保险理赔用的识别工具，而是智慧养殖工具。”系统研发方寿光市翔天农业科技有限公司副总经理刘守城说。每头牛都有“数字身份证”，全天候“点名”，可实现精确投喂。养殖数据直传农业、防疫部门，食品安全可追溯。试点以来，已覆盖超7000头牛，饲料利用率提升25%，发病率降低20%。","新华社\u002F兵团日报 2026-07-27","2026-07-27T00:00:00Z",60,{"impact":17,"substance":141,"depth":20,"authority":20,"freshness":69,"relevant":22,"comment":197},"报道了AI在奶牛养殖中的实际应用，有具体数据和案例，但时效性较差，且为地方性进展。",[199],{"name":193,"url":190},[27,201,28,29,202],"智慧大棚","寿光模式",{"code":60,"slug":204,"name":205,"path":206,"parent_code":207},"agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai","I",[],{"status":100,"locked":14,"version":35,"method":36,"evidence":210},{"candidates":211,"agricultural_evidence":42},[212,217,221],{"code":60,"score":213,"supported":42,"title":214,"abstract":215,"tags":216},9,[28],[],[28],{"code":47,"score":41,"supported":42,"title":218,"abstract":219,"tags":220},[],[51],[201],{"code":40,"score":41,"supported":42,"title":222,"abstract":223,"tags":224},[],[27],[27],[226,227],"农业人工智能 寿光模式 智慧养殖 智慧农业","农业人工智能 寿光模式","农业人工智能寿光模式智慧养殖智慧农业-126","2026-08-01T06:01:52.387430Z",{"id":231,"title":232,"url":233,"summary":234,"summary_zh":235,"content":8,"source_name":236,"source_url":233,"published_at":237,"category":138,"cover_url":8,"hotness":13,"is_selected":14,"score":238,"score_detail":239,"sources":242,"tags":244,"primary_topic":248,"secondary_topics":254,"classification":256,"search_phrases":273,"slug":276,"view_count":178,"doi":277,"paper":278,"created_at":293},5592,"Evaluation of Concept-Rank Collapse Audits for Computer Vision Agriculture Disease Classifiers Under Low-Label Regimes: An Interpretability Capacity Bound for Smallholder Fungicide Selectivity","https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.23545193","Concept-based interpretability methods for plant-disease classifiers are typically evaluated by whether their concept activations correlate with human-annotated labels, but this criterion is silent on a prior question: whether the classifier's internal representation retains enough concept-rank structure for any audit to be informative. the present analysis formalises that prior question as an interpretability capacity bound. A classifier is modelled as a concept-rank operator whose singular values govern how much concept-level signal survives the projection from pixel space to class logits, and a collapse audit is defined as the ratio of retained concept-rank energy to total concept energy. Under a low-label regime, where concept annotations are scarce and the audit must be estimated from few labelled examples, the bound predicts that audit reliability degrades as a function of the concept-rank spectrum and the label budget. The analysis is a theoretical and methodological contribution: it derives the bound, states the conditions under which a collapse audit is identifiable, and characterises the operating characteristics of the audit as a decision procedure under a declared variance structure (between-cluster SD = 6.0, residual SD = 4.5, intraclass correlation = 0.64). Power was estimated across 2000 design cells for cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points; 52 clusters were required for 5 percentage points at 80% power. The bound reframes concept audits as capacity tests rather than fidelity tests, with consequences for how smallholder fungicide-selectivity decisions should be supported by interpretable disease classifiers.","面向植物病害分类器的基于概念的可解释性方法，通常以其概念激活是否与人工标注标签相关作为评价标准，但该标准对一个更先前的问题保持沉默：分类器的内部表征是否保留了足够的概念秩结构，从而使任何审计具有信息量。本分析将该先前问题形式化为一个可解释性容量上界。分类器被建模为概念秩算子，其奇异值支配着从像素空间到类别logits的投影中概念级信号得以保留的程度，并将坍缩审计定义为保留的概念秩能量与总概念能量之比。在低标签情形下，概念标注稀缺且审计必须由少量标注样本估计，该上界预测审计可靠性会随概念秩谱和标签预算的函数而退化。本分析是一项理论和方法论贡献：推导出该上界，陈述坍缩审计可识别的条件，并在声明的方差结构（簇间SD = 6.0，残差SD = 4.5，组内相关系数 = 0.64）下刻画审计作为决策程序的操作特征。在2000个设计单元中，针对簇数为8、16、24和32以及效应量为3、5和8个百分点的情形估计了功效；在80%功效下达到5个百分点需要52个簇。该上界将概念审计重新框定为容量检验而非保真度检验，并对可解释病害分类器应如何支持小农杀菌剂选择性决策具有启示意义。","Zenodo (CERN European Organization for Nuclear Research)","2026-10-21T00:00:00Z",69,{"impact":143,"substance":240,"depth":17,"authority":20,"freshness":13,"relevant":22,"comment":241},21,"理论方法论文，提出可解释性容量上界并给出审计功效分析，对智慧农业病害识别可信度有参考价值，但属小众方法学研究、缺乏田间实证，暂不宜进入每日精选。",[243],{"name":236,"url":233},[27,28,245,246,247],"小农户","植物病害识别","可解释性",{"code":249,"slug":250,"name":251,"path":252,"parent_code":253},"G2","farm-services","社会化服务与规模经营","\u002Ftopic\u002Frural-services\u002Ffarm-services","G",[255],{"code":60,"slug":204,"name":205,"path":206,"parent_code":207},{"status":100,"locked":14,"version":35,"method":36,"evidence":257},{"candidates":258,"agricultural_evidence":42},[259,264,269],{"code":249,"score":20,"supported":42,"title":260,"abstract":262,"tags":263},[261],"smallholder",[261],[245],{"code":60,"score":13,"supported":42,"title":265,"abstract":267,"tags":268},[266],"computer vision",[],[28,246],{"code":40,"score":22,"supported":14,"title":270,"abstract":271,"tags":272},[],[],[27],[274,275],"农业人工智能 植物病害识别 可解释性 智慧农业","农业人工智能 植物病害识别","农业人工智能植物病害识别可解释性智慧农业-5592","10.5281\u002Fzenodo.23545193",{"doi":277,"openalex_id":279,"authors":280,"venue":236,"cited_by_count":178,"oa_url":233,"card":287,"direction":183,"ingested_from":292},"W7223137323",[281,283,285],{"name":282,"orcid":8},"Akshay Ramgoolam",{"name":284,"orcid":8},"Priya Bundhoo Appadoo",{"name":286,"orcid":8},"Jean-Noël LEspérance",{"tldr":288,"method":289,"finding":290,"direction":183,"opportunity":291},"提出可解释性容量上界，评估低标签下概念秩崩溃审计的可靠性。","将分类器建模为概念秩算子，推导容量上界并做2000格功效模拟。","审计可靠性随概念秩谱与标签预算下降；5个百分点效应需52个聚类达80%功效。","可将容量上界用于小农户杀菌剂选择性决策的可解释模型审计与标签预算设计。","openalex","2026-10-22T11:21:10.425290Z",{"id":295,"title":296,"url":297,"summary":298,"summary_zh":299,"content":8,"source_name":236,"source_url":297,"published_at":237,"category":138,"cover_url":8,"hotness":13,"is_selected":14,"score":238,"score_detail":300,"sources":302,"tags":304,"primary_topic":308,"secondary_topics":314,"classification":316,"search_phrases":341,"slug":344,"view_count":178,"doi":345,"paper":346,"created_at":358},5591,"Hierarchical Conformal Risk Control for Artificial Intelligence Agriculture Irrigation Scheduling Under Unmodeled Infiltration: A Computer Safety-Guarantee Evaluation","https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.23543665","Irrigation scheduling algorithms that rely on soil-moisture forecasts are vulnerable to unmodeled infiltration, the process by which water bypasses the root zone through preferential pathways that resistance-based soil models cannot represent. The framework is evaluated through a replication-based design study that characterises the operating characteristics of the before-after-control-impact estimator used to audit scheduler performance, under a declared variance structure with between-cluster standard deviation 6.0, residual standard deviation 4.5 and intraclass correlation 0.64. Power was estimated across 2000 design cells spanning cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points. The analysis indicates that 54 clusters are required to detect a 5-percentage-point effect at 80% power, and that the minimum detectable effect declines steeply as cluster count rises from 8 to 32. The contribution is a formal safety guarantee for irrigation decision support that remains valid when the infiltration model is wrong, together with an audit design that makes the guarantee empirically checkable.","依赖土壤湿度预报的灌溉调度算法容易受到未建模入渗的影响，即水分通过优先路径绕过根区的过程，而基于阻力的土壤模型无法表征这一过程。该框架通过一项基于重复的设计研究进行评估，该研究刻画了用于审计调度器性能的前后对照影响估计量的运行特性，其所依据的声明方差结构为：组间标准差6.0、残差标准差4.5、组内相关系数0.64。功效估计覆盖2000个设计单元，涵盖聚类数8、16、24和32以及效应量3、5和8个百分点。分析表明，在80%功效下检测5个百分点效应需要54个聚类，且最小可检测效应随聚类数从8增至32而急剧下降。本文的贡献在于为灌溉决策支持提供了一种正式的安全保证，该保证在入渗模型错误时仍然有效，同时提出了一种使该保证可经实证检验的审计设计。",{"impact":143,"substance":85,"depth":17,"authority":20,"freshness":213,"relevant":22,"comment":301},"该研究提出在入渗模型失配条件下仍有效的灌溉决策安全保证框架，并给出可审计的试验设计，方法新颖、结论可靠，但属细分方法学研究，产业影响有限，适合作为专业精选而非头条。",[303],{"name":236,"url":297},[27,28,305,306,307],"灌溉决策","安全保证","试验设计",{"code":309,"slug":310,"name":311,"path":312,"parent_code":313},"H3","irrigation","水资源与灌溉","\u002Ftopic\u002Fenvironment\u002Firrigation","H",[315],{"code":60,"slug":204,"name":205,"path":206,"parent_code":207},{"status":100,"locked":14,"version":35,"method":36,"evidence":317},{"candidates":318,"agricultural_evidence":42},[319,324,330,337],{"code":309,"score":19,"supported":42,"title":320,"abstract":321,"tags":323},[310],[322,310],"灌溉",[305],{"code":60,"score":20,"supported":42,"title":325,"abstract":327,"tags":329},[326],"artificial intelligence",[328],"decision support",[28],{"code":331,"score":72,"supported":42,"title":332,"abstract":333,"tags":336},"H2",[],[334,335],"土壤","soil",[],{"code":40,"score":22,"supported":14,"title":338,"abstract":339,"tags":340},[],[],[27],[342,343],"灌溉调度 安全保证","农业人工智能 安全保证 智慧农业 灌溉决策","灌溉调度安全保证-5591","10.5281\u002Fzenodo.23543665",{"doi":345,"openalex_id":347,"authors":348,"venue":236,"cited_by_count":178,"oa_url":297,"card":353,"direction":183,"ingested_from":292},"W7223131205",[349,351],{"name":350,"orcid":8},"Nassima Benkhelifa",{"name":352,"orcid":8},"Yacine Aït Ahmed",{"tldr":354,"method":355,"finding":356,"direction":183,"opportunity":357},"提出分层保形风险控制，为未建模入渗下的灌溉调度提供安全保证。","分层保形风险控制与基于重复设计的BACI审计，2000个设计单元功效分析。","检测5个百分点效应需54个集群，最小可检测效应随集群数增加陡降。","可将保形风险控制扩展到其他未建模扰动下的农业决策安全保证与审计设计。","2026-10-22T11:21:10.283561Z",{"id":360,"title":361,"url":362,"summary":363,"summary_zh":364,"content":8,"source_name":236,"source_url":362,"published_at":237,"category":138,"cover_url":8,"hotness":13,"is_selected":14,"score":365,"score_detail":366,"sources":368,"tags":370,"primary_topic":373,"secondary_topics":374,"classification":376,"search_phrases":391,"slug":394,"view_count":178,"doi":395,"paper":396,"created_at":410},5590,"Evaluation of Concept-Probe Temporal-Stability Audits for Computer Vision Agriculture Disease Classifiers Under Seasonal Phenology Drift: An Interpretability Decay Bound for Smallholder Fungicide Timing","https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.23545234","Concept-probe audits are increasingly used to certify that a crop-disease classifier's internal representations track the pathological concepts a smallholder would act on, but a probe certified at one point in the season may silently decouple from those concepts as phenology shifts the visual statistics of the field. the present analysis treats that decoupling as a formal object. It defines a temporal-stability audit as a sequence of concept-probe accuracies measured across defined phenological windows, and it derives an interpretability decay bound: the rate at which probe accuracy can fall before the fungicide-timing decision it is meant to support becomes unreliable. The bound is expressed as a function of the probe's initial margin, the drift in concept prevalence between windows, and the classifier's sensitivity to the drifted features, and it yields a certified audit interval rather than a single-point certificate. The paper then characterises the operating characteristics of the audit design under a declared variance structure, using a replication-based evaluation of the before-after-control-impact estimator with between-cluster SD = 6.0, residual SD = 4.5 and intraclass correlation = 0.64. Power was estimated across 2000 design cells over cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points; 54 clusters were required for 5 percentage points at 80% power. The contribution is a decision-theoretic account of when a concept-probe certificate expires, and a design rule for scheduling re-audits across the season.","概念探针审计（concept-probe audit）日益被用于验证作物病害分类器的内部表征是否追踪小农户会据以采取行动的病理学概念，但在生长季某一时点获得认证的探针，可能随着物候变化改变田间的视觉统计特征而悄然与这些概念脱钩。本分析将这种脱钩视为一个形式化对象。它将时间稳定性审计（temporal-stability audit）定义为在既定物候窗口内测量的一系列概念探针准确率，并推导出一个可解释性衰减界限（interpretability decay bound）：即探针准确率在其所支持的杀菌剂施用时机决策变得不可靠之前所能下降的速率。该界限表示为探针初始裕度、窗口间概念流行率漂移以及分类器对漂移特征敏感度的函数，并给出一个经认证的审计区间而非单点认证。随后，本文在声明的方差结构下刻画该审计设计的运行特征，采用基于重复的前后对照影响估计量评估，其中组间标准差 = 6.0、残差标准差 = 4.5、组内相关系数 = 0.64。功效在 2000 个设计单元上估计，聚类数取 8、16、24 和 32，效应量取 3、5 和 8 个百分点；在 80% 功效下达到 5 个百分点需要 54 个聚类。本文的贡献在于从决策论角度阐明概念探针认证何时失效，并给出在整个生长季内安排重新审计的设计规则。",70,{"impact":143,"substance":85,"depth":17,"authority":20,"freshness":13,"relevant":22,"comment":367},"该论文提出概念探针审计的时序稳定性框架与可解释性衰减界，方法新颖、统计设计扎实，但属小众方法论研究，产业影响有限，适合作为专业精选而非头条。",[369],{"name":236,"url":362},[27,28,245,371,372],"作物病害识别","模型审计",{"code":249,"slug":250,"name":251,"path":252,"parent_code":253},[375],{"code":60,"slug":204,"name":205,"path":206,"parent_code":207},{"status":100,"locked":14,"version":35,"method":36,"evidence":377},{"candidates":378,"agricultural_evidence":42},[379,383,387],{"code":249,"score":19,"supported":42,"title":380,"abstract":381,"tags":382},[261],[245,261],[245],{"code":60,"score":13,"supported":42,"title":384,"abstract":385,"tags":386},[266],[],[371,28],{"code":40,"score":22,"supported":14,"title":388,"abstract":389,"tags":390},[],[],[27],[392,393],"概念探针 时序稳定性 作物病害分类器","杀菌剂 施药时机 可解释性衰减","概念探针时序稳定性作物病害分类器-5590","10.5281\u002Fzenodo.23545234",{"doi":395,"openalex_id":397,"authors":398,"venue":236,"cited_by_count":178,"oa_url":362,"card":405,"direction":183,"ingested_from":292},"W7223088030",[399,401,403],{"name":400,"orcid":8},"Abdi-Basid Idriss Farah",{"name":402,"orcid":8},"Amina Abdillahi Guedi",{"name":404,"orcid":8},"Moussa Omar Youssouf",{"tldr":406,"method":407,"finding":408,"direction":183,"opportunity":409},"提出概念探针时间稳定性审计与可解释性衰减界，判断何时需重新审计。","定义跨物候窗口探针精度序列，推导衰减界，用BACI复制评估统计功效。","5个百分点效应需54个聚类达80%功效，审计证书会随物候漂移过期。","可将衰减界与再审计调度规则嵌入小农户杀菌剂时机决策支持系统，并验证跨作物泛化。","2026-10-22T11:21:08.978348Z"]