继续创作，加快生长！这是我参与「日新计划 6 月更文挑战」的第22天，点击检查活动概况

ShowMeAI日报系列全新晋级！掩盖AI人工智能 东西&ampmysql安装配置教程;结构 | 项目&代码 | 博文&共享 | 数据&ampmysql数据库命令大全;资源 | 研究&论文 等方向。点击检查 历史文章列表，在大众号内订阅论题 #ShowMeAI资讯日报，可接收每日最新推送。点击 专题合辑&电子月刊 快速阅读各专题全集。

1.东西&结构

东西：Visual Studio Code的DV开源C扩展，用于版本办理、试验办理、数据流程可视化

‘DVC Extension for V梯度下降法isual Studio Code – Machine learning experiment tracking and data versioning with DVC extension for VS Code’ by Iterative

GitHub: gimysql数据库命令大全thub.com/itera架构图怎么制作tive/v…

东mysql密码忘记了怎么办西渠道：Mage – 开源数据办理渠道，可用来清洗数据，为练习AI/ML模型做准备

‘Mage – Mlinux系统安装age is alinux创建文件n open-source data management platform that helps you clean datmysql面试题a a架构图模板nd prepare it for training AI/ML models.’

GitHub: github.com/mage-ai/mag…

东西结构：Torch Poi架构nts3D – 点云深度学习统一开源软件结构

《To架构师和程序员的区别rch Points3D — A unifying framework for deep learning on point clouds》by Nicolas Chaulet

GitHub: github.com/nicolas-cha…

东西库：torchsnapshot – 用于为PyTorch大规划散布式练习工作负载供给容错才能的轻量库

‘torchsnapshot – A light-weight library for adding fault tolerance to large-scale PyTorch distributed training w架构图怎么制作orkloads.’ by Meta Research

GitHublinux系统安装: github.com/facebookres…

东西库：freemocapmysql怎么读 – 3D人体骨架检测

‘freemocap –架构 Free like Freedom’ by jonmatthis

GitHub: github.com/jonmatthis/…

东西：Beekeeper Studio – 一linux常用命令款开源的跨渠道 SQL 编辑器

Beekeeper Studio供给 SQL 语法高亮、自动补全、数据表内容挑选与过滤、衔接 Web 数据库、存储历史查询记载等功用。

支持 SQLite、MySQL、MariaDB、Postgres 等主流数据库，并兼容 Windows、macOS、Linux 等桌架构师工资面操作体系。

GitHub: github.com/beekeeper-s…

2.博文&共享

免费书本：南瓜书PumpkinBook架构师证书

地址: datawhalechina.git开源软件hub.io/pumpkin-boo…

“周志华老师的《机器学习》（西瓜书）是机器学习范畴的经典入门教材之一，周老师为了使尽可能多的读者经过西瓜书对机器学习有所了解, 所以在书中对部分公式的推导细节没有详述，可是这对那些想深究公式推导细节的读者来说可能“不太友好”，本书旨在对西瓜书里比较难了解的公式加以解析，以及对部分公式弥补详细的推导细节。”

3.数据&资源

资源列表：计算机视觉最佳实践、代码开源是什么意思示例和相关文档

‘Computer Vision – Best梯度 Practices, code samples, and documentation for Computer Vision.’ by Microsoft

GitHub: github.com/microso梯度稀释的目的ft/c…

课程资料：STAT 991 – Topics In Modern Statistical Learning (UPenn, 2022 Spring)：UPenn现代计算学开源众包习专题课程资料

‘STAT 991: Topics In Modern Statistical Learning (UPenn, 2022 Spring) – focus on uncertainty qua架构图怎么画ntification’ by Edga架构是什么意思r Dobriban

GitHub: github.com/dobriban/To…

资源列表架构图模板：深度学习图反常检测文献资源列表

‘Awesome-Deep-Graph-Anomaly-D梯度公式etection – Awesome graph anomaly detection techniques built based on deep learning frameworks. ‘mysql基础命令 by XiaoxiaoMa-MQ

GitHub: github.com/XiaoxiaoM架构a-…

4.研究&论文

大众号回复关键字 日报，免费获取整理好的6月论文合辑。

论文：ReCo: Retrieve and Co-segment for Zlinuxero-shot T架构师ransfer

论文标题：ReCo: R架构工程师etrimysql索引eve and Co-segment for Zero-shot Transfer

论文时刻：14 Jun 2022

所属范畴：计算机视觉

对应使命：Semantic Segmentation，Unsu梯度稀释pervised Semantic Segmentation，语义切割，无监督开源节流是什么意思语义切割

论文地址：arxiv.or开源节流g/abs/22梯度下降06.07…

代码完成：github.com/NoelShin/re…

论文作者：Gyungin Shin, Weidi Xie, Samuel Albanie

论文简介：Semantic segmenta开源阅读tlinux必学的60个命令ion has a broad range of appli梯度下降法原理cation架构图怎么画s, but its r开源节流是什么意思eal-world impact has been significantly limited by the prohibitive annotation costs necessary to enable deployment. / 语义切割具有广泛的运用，但其对实际国际的运用布置受到所需的高昂注释本钱明显限制。

论文摘要：Semantic梯度公式 segmentation has a broad range of applica架构师证书tions, but its realinux是什么操作系统l-world impact has been significamysql安装ntly limited by the prohibitive annotatmysql怎么读ion costs nece架构师工资ssary to enable deployme架构师nt. Segmentation methods that forgo supervision can side-step these costs, but exhibit the inconvenient requirement to provide labelled examples from the target distribution to assign concept names to predictions. Anmysql数据库命令大全 alternative line of work in language-linux系统安装imlinux必学的60个命令age pre-training has recmysql索引ently demonstrated the potential to梯度稀释 produce models that can both assign names ac梯度怎么求ross large vocabularies of concep开源阅读ts and enable zero-shot transfer for classification, but do not demonstrate commensurate segmentation abilitie开源节流s. In this work, we strive to achieve a synthesis of these two approaches that combmysql安装ines their strengths. We leverage the retrieval a梯度稀释的目的bilities of one such language-image pre-trained model, CLIP, to dynamically curate training sets from unlabelled images for arbitrary collections of concept na开源软件mes, and leverage the robust correspondences offered by modern image representations to co-segment entities among the resulting collections. The synlinux虚拟机thetic架构图怎么制作 segment collections are then employed to construct a segmentatmysql基础命令ion model (without梯度的几何意义 requiring pixel labels) whose knowl梯度下降法原理edge of concepts is inhmysql数据库erited from the scalable pre-training procesmysql面试题s of CLIP. We demonstrate that olinux系统ur approach, termed Retrieve and Co-segment (ReCo) performs favourably to unsupervised segmentation approaches while inheriting the conv梯度enience of nameable predictions and zero-shot transfer. We also demonstrate ReCo梯度的几何意义‘s ability to generate specialist segmenters for extremely rare objects.

语义切割具有广泛的运用，但其实梯度的几何意义际国际的运用布置mysql基础命令受到所需的高昂注梯度释本钱明显限制。抛弃监架构图督的切割办法能够逃避这些本钱，但表现出不方便的要求，即供给来自目标散布的符号示例以将概念名称分配给猜测。言语-图画预练习的另一项工作最近证明了生成模型的潜力，这些模型既能linux常用命令够跨大型概念词汇表分配名称，又能够完成零样本搬迁以进行分类，但不能表现出相应的切割才能。在这项工作中linux命令，咱们努力完成这两种办法的结合，综合它们的优势。咱们运用这样mysql索引一种言语图画预练习模型 CLIP 的检索才能，从未符号图画中为恣意概念名称调集动态办理练习集，并运用现架构是什么意思代图画标明供给的鲁棒对应联系来一起切割实开源中国体之间的实体。然后运用组成的分段调集来构建分段模型（不需求像素标签），其概念常梯度是什么意思识承继自 CLIP 的可扩展预练习进程。咱们证明了咱们的办法，称为检索和协linux常用命令同切割（ReCo），优于无监督切割办法，同时承继了可命名猜测和零样本搬迁的便利性。咱们还展现了 ReCo 极其稀有的物体生成专业切割器的才能。

论文：DoWhy-GCM: An extension of DoWhy for causal inference in graphical causal models

论文标题：Domysql索引Why-GCM: An e开源节流是什么意思xtension of DoWhy for causal infermysql数据库ence in graphical causal models

论文时刻：14 Jun 2022

所属范畴：归因剖析

对应使命：归因剖析

论文地址：arxiv.org/abs/2206.06…

代码完成：g开源节流ithub.com/py-why/dowh…

论文作者：Patrick Blbaum, Peter Gtz, Kailash Budhathoki, Atalanti A. Mastakouri, Dominik Janzing

论mysql密码忘记了怎么办文简介：We introduce DoWhy-GCM, an extension of the DoWhy Python library, that leverages graphical causal models. / 咱们带来了Do架构Why-GCM，它是 DoWlinux是什么操作系统hy Python 库的扩架构是什么意思展，它运用了图linux是什么操作系统因果模型。

论文摘要：We introduce DoWhy-GCM, an extension of the DoWhy Python library, that leverages graphicallinux causal models. Unlike existing causality libraries, which mainly focus on effect estmysql数据库imation questions, with DoWhy-GCM, users can ask a wide range of additional camysql怎么读usal questions, suchlinux虚拟机 as identifying the root causes of outliers and dimysql索引stributional changes, causal structure learning, attributing causal influences, and diagnosis of causal structures. To this end, DoWhy-GCM users first model cause-effect relations between variables in a system under study through a grap梯度hical causal model, fiMySQLt the causal mechanisms of variables next, and then ask the causal question. All these steps take only a few li梯度是什么意思nes of code in DoWhy-GCM. The library is avail梯度是什么意思abllinux系统安装e at github.com/py-why/dowh…

咱们带来了 DoWhy-GCM，它是 DoWhy Python 库的扩展，它运用了mysql安装配置教程图因果模型。 与主要关注作用估量问题的现有因果联系库不同，运linux系统用 DoWhy-GCM，用户能够提出广泛的附加因果问题，例如识别反常值和架构散布改变的根本原因、因果开源矿工结构学习、归因因果影响以及因果结构的确诊。 为此，DoWhy-GCM 用户首要经过图因果模型对所研linux虚拟机究体系中变量之间的因果联系进行建模，然后拟合变量的因果机制，然后提出因果问题。 所有这些进程在linux命令 DoWhy-GCM 中只需求几行代码。 该库坐落 github.com/py-why/dowhlinux必学的60个命令…

论文：LST:架构师和程序员的区别 Ladder Side-Tuning for Parameter and Memory Efficient Transfer Learning

论文开源阅读标题：开源LST: L架构图模板adder Side-Tuning for Parameter and Memory Efficient Transfer Learning

论文时刻：13mysql数据库命令大全 Jun 2linux命令022

所属范畴：计算机视觉，自然言语处理

对应架构使命：Transfer Learning，Visual Question Answering，VQA，搬迁学习，视觉问答

论文地址：arxiv.org/abs/2206.06…

代码完成：linux常用命令github.com/ylsung/ladd…

论文作者：Yi-Lin Sung, Jaemin Cho, Momysql安装hit Bansal

论文简介：LST saves 69% of thlinuxe memory costs to fine-tuLinuxne the whole network, while other me架构师和程序员的区别thods only save 26% of that in similar pa开源阅读app下载安装rameter usages (hence, 2. 7x more memory savings). / LST 为微调整个网络节省了 69%Linux 的内存本钱，而其他办法在相似的参数运用情况下仅节省了 26%（因此，节省了mysql安装配置教程 2. 7 倍的内存）。

论文摘要：linux系统Fine-tuning lmysql索引arge pre-trained models on downstream tasks has been adopted in a variety of dmysql数据库omains recently. However, it is costly to update the entire parameter set of large pre-trained models. Although recently proposed parameter-efficient transfer learning (PETL) techniques allow updating a small subset of parameters (e.g. only using 2% of para梯度洗脱meters) inside a pre-trained梯度稀释 backbone network for a new task, they only reduce the training memory requiremelinux重启命令nt by up to 30%. This is because the gradient computation for the trainable parmysql索引ameters still requires backpropag架构工程师atiomysql安装n through the large pre-架构图模板trmysql安装ained backbone model. To address t开源节流是什么意思his, we propose Ladder Side-Tu梯度怎么求ning (LST), a new PETL technique that reduces training memory requimysql安装配置教程rements by more substantial amounts. Unlike existing parameter-efficient method架构师证书s that insert additional parameters inside backbone networks, we train a ladder side network, a small and separate network that takes intermmysql密码忘记了怎么办ediate activ架构图模板ations as input via shortcut connections (ladders) from backbone networks an开源众包d makes predictions. LST has significantly lower memory requiremelinux必学的60个命令nts than previous methods, because it does not require ba开源节流是什么意思ckpropagation through the backbone network, but instead only through the side network and ladder connections. We eva开源阅读app下载安装luate our method with various models (T5, CLIP-T5) o开源软件n both NLP (mysql面试题GLUE) and vision-language (VQA架构, GQA, NLVR2, MSCOCO) tasks. LST saves 69% of the memory costs to fine-tune the whole network, while other methods only save 26% of that in开源节流是什么意思 similar parameter usages (hence, 2.7x more memory savings). Moreover, LST achieves higher accuracy than Adapter and LoRA in a low-memory regime. To further show the advantage of this better memory efmysql索引ficiency, we also apply LST to larger架构师证书 T5 molinux系统安装dels (T5-large, T5-3B), attaining bett梯度下降法原理er GLUE perfor梯度mance than fmysql怎么读ull fine-tunin架构师证书g and other PETL methods. The梯度下降法原理 exact same trend also holds in our experimenmysql数据库基础知识ts on VL tasks.

最近，各种范畴都选用了对下流使命的大开源节流是什么意思型预练习模型进行微调。可是，更新大型预练习模型的整个参数集的本钱很高。虽然最近提出的参数高效搬迁学习 (PETL) 技能答应在预练习的主干网络中为新使命更新一小部分参数（例如，仅运用 2% 的参数），但它们最多只能将练习内存需求削减30%。这是由于可练习参数的梯度计算依然需开源阅读app下载安装求经过大型预练习主干linux模开源节流型进行反向传达。为了处理这个问题，咱们提linux出了 Ladder Side-Tuning (LST)，这是一种新的 PE开源阅读app下载安装TL 技能，可大幅削减练习内存需求。与现有的在主干网络中插入额定参数的参数有用办法不同，咱们练梯度怎么求习了一个梯形侧网络，这是一个小型且独立的网络，它经过来自主干网络的方便衔接（“梯子”跳接）将中心激活作为输入并进行猜测。 LST 比曾经的办法具有明显下降的内存需求，由于它不需求经过主干网络进行反向传达，而只需求经过侧网络和跳跃“梯子”衔接。咱们在 NLP (GLUE) 和视觉言语 (VQA, GQA, NLVR2, MSCOCO) 使命上运用各种模型 (T5, CLIP-梯度洗脱T5) 评价咱们的办法。 LST 为微mysql安装调整个网络节省了 69% 的内存本钱，而其他办法在相似的参数运用情况下仅节省了 26%（因此，节省了 2.7 倍的内存）。此外，LST 在低内存状态下完成了比 Adapter 和 LoRA 更高的精度。为了进一步展现这种更好的内存效率的优势，咱们还将 LST 运用于更大的 T5 模型（T5-larg开源节流e，T5-3B），取得比彻底微调和其他 PETL 办法更好的 GLU开源节流是什么意思E 功用。在咱们对 VL 使命的试验中也存在彻底相同的趋势。

论文：Neural Prompt Search

论文标题：Neural Prompt Search

论文时刻：9 Jun 2022

所属范畴：计算机视觉

对应使命：Few-Shot Learning，Neural Architecture Search，Prompt Enginee梯度是什么意思ring，Transfer Learning，小样本学习、神经架构查找、搬迁学习

论文地址：arxiv.omysql基础命令rg/abs/2206.04…

代码完成：github.com/Davidzhangy…

论文作者：Yuanhan Zhang, Kaiyang Zhou, Ziwei Liu

论文简介：The size of vision models has grown exponentially over the last few架构是什么意思 years, especially after the emergence of Vision Transformer. / 视觉模型的规划在曩昔几年呈指数级增加，尤其是在 Vision Transformer 呈现之后。

论文摘要：The size of vision models has grown exponentially over themysql基础命令 last few years, especially after the emergence of Vision Transformer. This has motivated the develomysql索引pment of parameter-efficient tuning methods, such as learning adapter layers or visual prompt tokens, which allow a tiny portion of model parameters to be trained whereas梯度 the vast majority obtained from pre-training are frozen. However, designing a proper tuning method is non-trivial: one might need to try out a lengthy list of dlinux重启命令esign choiclinux常用命令es, not to mention that each downstrea开源矿工m dataset often requires custom designs. In this paper, we view the existing parameter-efficient tuning meth架构师ods as “prompt modules”开源节流 and p梯度公式ropose Neural prOmpt seArcH (NO梯度的几何意义AH), a novel approach that learns, for large vision models, the optimal design of prompt modules through a neural architecture search algorithm, sp开源代码网站githubecifically for each downstream dataset. By colinux虚拟机nducting extensive experiments on o梯度稀释的目的ver 20 vision datasets, we demonstrate that NOAH (i) is superior to individual prompt modules, (ii) has a good few-shot learning a架构图怎么制作bility, and (iii) is domain-generalizable. The code and models are available at github.com/Davidzhamysql索引ngy…

视觉模型的规划在曩昔几年呈指数级增加，尤其是在MySQL Vision Transformer 呈现之后。这推动了参数高效调整办法的开发，例如学习适配器层或视觉梯度稀释的目的提示符号，它们答应只练习一小部分模型参数，而从预练习中取得的绝大多数参数被冻住。但是，规划适宜的调优办法并非易事：可能需求测验一长串规划梯度稀释挑选，更不用说每个下流数据集一般都需求定制规划梯度下降。在本文中，咱们将现有的参数有用调整办法视为“提示模块”，并提出了神经提示查找（架构图怎么制作NOAH），这是一种新办法，能够经过神经架构查找算法学习大型开源阅读视觉模型的提示模块的优linux重启命令化规划，特别是针对每个下流数据集。经过对 20 多个视觉数据集进行广泛的试验，咱们证明 NOAH (i) 优于单个提示模块，(ii) 具有良好的小样本学习才能，以及 (iii) 是域泛化的。代码和模型可在 github.co架构是什么意思m/Da梯度公式vidzhangy… 取得。

论文：RelVi开源中国T: Concept-guided Vision Transformer for Visual Relational Reasoning

论文标题：RelViT: Concept-guided Vision Transformermysql安装配置教程 for Visual Re架构图lational Reasoning

论文时刻：ICLmysql基础命令R 2022

所属范畴：计算机视觉

对应使命：Hum开源中国an-Object开源众包 Inter梯度稀释actiomysql怎么读n Detection，开源节流Systematic Generalization，Visual QuestioLinuxn Answering，Visual Reasoning，Zero-S梯度hot Human-Object Interaction Detection，人机交互检测、体系泛化、视觉问答linux创建文件、视觉推理、零样本梯度下降法人机交互检测

论文地址：arxiv.org梯度的几何意义/abs/2204.11…

代码完成：github.com/NVlabs/RelV…

论文作者：Xiaojian Ma, Weili Nie, Zhiding Yu, Huaizu Jiang, Chaowei Xiao, Yuke Zhu, Song-Chun Zhu, A梯度下降法nima Anandkumar

论文简介：This task remains challen开源是什么意思ging for current deep learning algorithms since it requires addressing three key technical prolinux系统blems jointly: 1) identifying object entities and their properties, 2)mysql面试题 inferri开源矿工ng semantic relations between pairs of entities, and 3) generalizing to novel object-relation combinations, i. e., sylinux创建文件stematic generamysql数据库lization. / 这项使命关于开源阅读app下载安装当时的mysql面试题深度学习算法依然具有挑战性，由于它需求一起处架构图怎么画理三个关键技能问题：1）识别目标实体及其特点，2）推断实体对之间的语义联系，以及 3）推广架构师证书到新的目标联系组合，也即“体系性泛化”。

论文摘linux重启命令要：Reasoning about visual relationshi架构师ps is central to how humans interpret the visual world. This梯度的几何意义 task remains challenging for current deep learning algorithms since it requires addressing three key technical problems jointly: 1) identifying object entities and their properties, 2) inferring semantic relations between pairs梯度下降法 of entities, and 3) generalizing to novel object-relation combinations, i.e., systematic genera开源lization. In this work, we use vision transformers (ViTs) as our base model for visual reasoning and make betlinuxter use of concep梯度公式ts defined as obje开源阅读ct entities and their relinux重启命令lations to improve the reasoning ability of ViTs. Specifical梯度洗脱ly, we introduce a novel conce梯度稀释pt-fea架构师ture dictionary to allow flexible image feature retrieval at training time with concept keys. This dictionary enables two ne架构师证书w concept-guided auxiliary tasks: 1) a global task for promoting relational reasoning, and 2) a local task for facilitating semantic object-c架构图怎么制作entric correspondence learning. To examine the systematic generalization of visual reasoning models, we introduce systematic splits for the standard HICO and GQA benchmarks. We show the resulting model, Concept架构师工资-guided Vision Transformer (or RelViT for short) siglinuxnificantly outperforms prior approaches on HIC架构图怎么制作O and GQA by 16% and 13% in the original slinuxplit, and by 43% and 18% in the systematic split. Our ablation an架构图怎么制作alyses aLinuxlso reveal our model’s compatibility with multiple ViT vmysql安装配置教程ariants and架构图怎么制作 robustness to hyper-parameters.

关于视觉联系的推理是人类如何解说视觉国际的中心。这项使命关于当时的深度学习算法依然具有挑战性，由于它需求开源一起处理三个关键技能问题：1）识别目标实体及其特点，2）推断实体对之间的语义联系，以及 3）推广到新的目标联系组合，即，体系性泛化。在这项工作中，咱们运用视觉transformers (ViTs) 作为视觉推理的根底模型，并更好地运用界说为目标实体及其联系的概念来进步 ViTs 的推理才能。详细来说，咱们引入了一种新颖的概念特征字典，以答应在练习时运用概念键进行灵活的图画特征检索。该词典支linux是什么操作系统持两个新的概念引导辅佐使命：1）促进联系推理的全局使命，以及 2）促进以语义目标为中心的对应学习的部分使命。为了检查视觉推理模型的体系归纳，咱们为规范 HICO 和 Glinux系统QA 基准引入了体系拆分。咱们展现了由此发生的模型，Concept-guided Vision Transformer（或简称 RelViT）在原始切割中明显优于 HImysql怎么读CO 和 GQA 的先前办法 16% 和 13%架构图，在体系切割中别离优于 43% 和 18%。咱们的融化剖析还揭示linux重启命令了咱们的模型与多种 ViT 变体的兼容性以及对超参数的鲁棒性。

论文：Optimal Transport Tools (OTT): A JAX Toolbox for all things Wasserstein

论文标题：Optimal Tralinux系统nsport Tools (OTT): A JAX Toolbox for all things Wasserstein

论文时刻：28 Jan 2022

论文地址：arxiv.org/abs/2201.12…

代码完成：github.com/ott-jax/ott

论文作者：Marco Cuturi, Laetitia Meng-Papaxan梯度是什么意思thos, Yingtao Tia梯度的几何意义n, Charlotte Bunne, Geoff Davis, Olivier Teboul

论文简介linux是什么操作系统：Optimal transport tools (OTT-JAX) is a Python toolbox that can solve optimal transport problems between point clouds and histograms. / 最优传输东西（OTT-JAX）是一个 Pythmysql基础命令on 东西箱，能Linux够处理点云和直方图之间的最优传输问题。

论文摘要：Optimal transpolinux创建文件rt toomysql数据库命令大全ls (OTT-JAX) is a Python toolbox that can solve optimal transport problems between point clouds and histograms. The toolbox builds on variLinuxous JAX features, such as automatmysql基础命令ic anmysql基础命令d custom reverse mode differentiation, vectorization, just-in-time compilation and accelerators support. The toolbox covers elelinux必学的60个命令mentary computations, such as the resolution of the regularized OT problem, and morelinux是什么操作系统 advanced extensi架构师和程序员的区别ons, such as barycmysql安装配置教程enters, Gromov-Wasserstein, lo梯度w-rank solvers, estimation of convex maps, dif开源阅读ferentiable generalizations o梯度的几何意义f quantiles and ranks, and approxi梯度稀释的目的mate OT between Gaussian mixtures. The toolbox code架构师证书 i开源s available at github.com/otlinux必学的60个命令t-jax/ott

最优传输东西（OTT-JAX）是一个 Python 东西箱，能够处理点云和直方图之linux是什么操作系统间的最优传输问题。 该东西箱建立在各种 JAX 功用之上，例如自动和自界说反向形式区分、矢量化、即时编译和加快器支持。 该东西箱包括基本计算，例如正则化 OT 问题的处理，以及更高级的扩展，例如重心、Gromov-Wasserstein、低秩求解器、凸图估量、分位数和秩的可微泛化以及之间的近似 OT 高斯混合。 东西箱代码能够在 github.com/ott-jax/ott 取到

论文：SaRNet: A Dataset for Deep Lelinux常用命令arning Assisted Se开源阅读app下载安装arch and Rescue with Satellite Imagery

论文标题：SaRNet: A Dataset for Deep Learning Assisted Search and Rescue with Satellite Imagery

论文时刻：26 Jul 2021

所属范畴：计算机视觉

对应使命：Humanitarian，object-detection，Object Detection，人道主义，物体检测，物体检测

论文地址：arxiv.org/abs/2107.12…

代码完成：github.com/michaelthor…

论架构师和程序员的区别文作者：Michael Thoreau, Frazer Wilson

论文简介：Access to high resolution satellite imagery h架构师证书as dralinux系统matically incmysql数据库reased in recenlinux系统安装t ye开源众包ars as several new conste开源是什么意思llations have entered service. / 近年来，跟着几个新卫星的投入运用，高分辨率卫星图画的获取明显增加。

论文摘要：Access to high resolution satellite imagery has dramatically increased in recenlinux重启命令t years as several new constellations have enter开源众包ed service. High revisit frequencies as welllinux as improved resolution has widened the use cases of satellite ilinux虚拟机magery to ar开源是什么意思eas such as humanitarian relief and even Search and Rescue (SaR). We propose a novel remote sensing object detection dataset for deep learning assisted SaR. This datalinux虚拟机set contai架构师工资ns only small objects that ha开源软件ve been identified as pot开源阅读app下载安装ential targets as part of a live SaR response. We evaluate the applicati开源代码网站githubon of popular object detection models to this dataset as a baseline to inform further research. We also propose a novel object detection metric, specifically designed to be used in a deep learning assisted SaR setting.

近年来，跟架构师着几个新卫星的投入运用，对高分辨率卫星图画的拜访急剧增加。 高重访频率和更高的分辨率已将卫星图画的用例扩大到人道主义救济乃至搜救 (SaR) 等范畴。 咱们提出了一种用于深度学习辅佐 SaR 的新型遥感目标检测数据集。 该数据集仅包含作为实时 SaR 呼应linux常用命令的一部分被识别为潜在目标的小目标。 咱们评价流行的目标检测模型在该数据集上的架构图怎么画运用，作为进一步研究的linux是什么操作系统基准。 咱们还提出了一种新颖的目标检测目标，专门规划用于深度学梯度下降法原理习辅佐 S开源软件aR 的场景。

论文：Direct Voxel Grid Optimization: Super-fast Convergence for Radiance Fields Reconstruction

论文标题：Direct Voxel Grid Optimization: Super-flinux创建文件a梯度稀释的目的st Convergence for Radiance Fimysql数据库命令大全elds Reconstruction

论文时刻：CVPR 2022

所属范畴：计算机视觉

对应使命：Novel View Synthesis，新视图组成

论文地址：arxiv.org/abs/2111.11…

代码完成：github.com/sunset1995/…

论文作者：Cheng Sun, Min Sun, Hwann-Tzong Chen梯度的几何意义

论文简介：Flinux必学的60个命令inally, evaluation on five inward-facing benchmarks shows that our method matches, if not surpassesmysql索引, NeRF’s quality, yet it only takes about 15 minutes to train from scratch for a new scene. / 最终，对五个内向基准的评价标明，咱们的办法与 NeRF 的质量相匹配，乃架构工程师至超越，但从头开始练习新场景只需求大约 15 分钟。

论文摘要：We present a super-fast convergence approach to reconstructi架构工程师ng the per-smysql安装cenelinux系统安装 r开源众包adian梯度下降ce field from a set of images that capture th开源软件e scene with known poses. This task, which is often applie架构师d to novel view synthesis, is recently revolutionized by Neural Radianlinux系统ce Fieldmysql索引 (NeRF) for its state-of-the-art quality and架构图模板 flexibility. Ho梯度下降法wever, NeRF and its variants require a lengthy training time架构图怎么制作 ranmysql数据库基础知识ging from hours to days for a single scene. In contrast, our approach achieves NeRF-comparable qual开源矿工ity and converges rapidly from scratch in less than 15 minutes with a single GPU. We adopt a representation consistin架构图怎么制作g of a density voxel grid for scene geometry and a feature voxel grid with a shallow network for complex view-dependent appearance. Modeling with explicit and discretized volume represlinux创建文件entations is not new, but we propose two simple yet non-trivial techniques that contri架构图怎么画bute to fast convergence speed a梯度下降法原理nd high-qumysql索引ality output. First, we i架构师ntroduce the post-activation interlinux必学的60个命令polation on voxel densi架构工程师ty, which is caplinux创建文件able of produc开源中国ing sharp surfaces in开源阅读 lowlinux是什么操作系统er grid resolution. Second, direct voxel density op架构图怎么画timization i梯度洗脱s prone to suboptimal geometry solutions, so we robustify the optimiz开源阅读ation process by imposing several prior梯度稀释s. Finally, evaluation on five inward-facing bench架构师证书marks shows that our method matches, if not surlinux常用命令passes, NeRF’s quality, yet it only takes about 15 minutes to train from scratchmysql数据库 fo梯度公式r a new scene.

咱们提出了一种超快开源节流速收敛办法，用于从一组捕获具有已知姿态的场景的图画中重建每个场景的辐射场。这项使命一般运用于新颖的视图组成，最近因其最先进的质量和灵活性而被神经辐射场 (NeRF) 彻底改变。但是，关于单个场景，NeRF 及其变体需求很长的练习时刻，从数小时到数天不等。相比之下，咱们的办mysql基础命令法完成了与 NeRF 相当的质量，并在不到 15 分钟的时刻内仅凭单个 GPU 从头练习能够完成快速收敛。咱们选用由用于场景几何的密度体素网格和具有浅层网络的特征体素网格组成的标明，用于复杂的依赖于视图的外观。运用显开源矿工式和离散化的体积标明进行建模并不新鲜，但咱们提出了两种linux必学的60个命令简略但非有用的技能，有助于快速梯度下降法原理收敛和高质量输出。首要，咱们介绍了体素密度的激活后插值，它能梯度下降够以较低的网格分辨率发生锋利的外表。其次，直接体素密度优化简开源矿工单呈现次优几何处理方案，因此咱们经过强加几个先验来加强优化进程。最终，对五个内向基准的评价标明，咱们的办法与 NeRF 的质量相匹配，梯度怎么求乃至超越，但从头开始练习新场景只需求大约 15 分钟。

咱们是 Sho架构图怎么制作wMeA架构师证书I，致梯度公式力于传达AI优质梯度稀释的目的内容，共享行业处理方案，用常识加快每一次技能生长！点击检查 历史文章列表，在大众号内订阅论题 #ShowMeAI资讯日报，可接收mysql面试题每日最新推送。点击 专题合辑&电子月刊 快速阅读各专题全集。

• 作者：韩信子@ShowMeAI
• 历史文章列表
• 专题合辑&电子月刊
• 声明：版权所有，转载请联系渠道与作者并注明出处
• 欢迎回复，托付点赞，留言引荐中有价值的文章、东西或主张，咱们都会尽快回复哒~