贵在坚持

文章目录Multimodal Transformer for Unaligned Multimodal Language SequencesSummary摘要 (中文)Research ObjectiveBackground and Problemsmain workRelated workMethod(s)ConclusionReference(optional)Arouse for meMultimodal Transformer for Unaligned Multimodal Language

文章目录backgroundOKR1. Unimodal basic representations(visual, language,speech )visual2. language3. audio2. Data-Driven Machine Learning1. K-Nearest Neighbor3. Linear Classification1. Interpreting Multiple Linear Classifiers： ![在这里插入图片描述](https://img-blog.cs

文章目录backgroundOKR1. Components of a neural network2. Learning the model3. Convolutional Neural networksbackgroundRecently, I find a good cources about multimodal machine learning. In this blog, I will study it and note my understanding.Here is orgin URL

文章目录backgroundOKR1. Components of a neural networkbackgroundRecently, I find a good cources about multimodal machine learning. In this blog, I will study it and note my understanding.Here is orgin URL : pptOmaster basic knowledge about CNN architectu

文章目录backgroundOKR1.what is modality ?backgroundRecently, I find a good cources about multimodal machine learning. In this blog, I will study it and note my understanding.Omaster multimodal base workKRwhat is modality ?multimodal develop historyma

backgroundRecently, I find a good cources about multimodal machine learning. In this blog, I will study it and note my understanding.Omaster multimodal base workKRwhat is modality ?multimodal develop historymain area in multimodal3. main areas i

文章目录backgroundOKR1. Unsupervised representation learning2. Learning the model3. Convolutional Neural networksbackgroundRecently, I find a good cources about multimodal machine learning. In this blog, I will study it and note my understanding.Here is org

文章目录backgroundOKR1. Components of a neural network2. Learning the model3. Convolutional Neural networksbackgroundRecently, I find a good cources about multimodal machine learning. In this blog, I will study it and note my understanding.Here is orgin URL

文章目录backgroundOKR1. Quick recapbackgroundRecently, I find a good cources about multimodal machine learning. In this blog, I will study it and note my understanding.Here is orgin URL : pptOhow to set coordinated representations.KRQuick recapCoordin

Disjoint Mapping Network for Cross-modal Matching of Voices and FacesSummaryIn this paper, author propose DIMNets, a framework that formulates the problem of cross-modal matching of voices and faces. It is not map voices to faces directly.In this fram

Multimodal data fusion framework based on autoencoders for top-N recommender systemsSummaryIn this paper, author design a new framework based on autoencoders to improve video’s recommendation quality. This method effection comfirm by写完笔记之后最后填，概述文章的内容，以

Opeer review 需要包含的内容（不要留情，提交之前你对朋友的文章留情了，审稿人会留情？）整个过程可以用Word的评审模式，也可以直接提交markdown。有的人写论文时用latex,建议使用latex转Word后再交给同学评审。KR整体评价看不懂的地方有哪些？语法建议其他补充1. 整体评价（写在论文题目附近）1. 创新点评价（如果对这个方向没了解，可以不用写）2. 文章可读性评价（必须要写奥）3. 补充（补充一些其他的评价）2. 单个章节评价(单个章节评价是基

文章目录Multimodal data fusion framework based on autoencoders for top-N recommender systemsSummaryResearch ObjectiveBackground and ProblemsRelated workMethod(s)EvaluationConclusionReference(optional)Arouse for meMultimodal data fusion framework based on aut

文章目录On Learning Associations of Faces and VoicesSummary摘要 (中文)Research ObjectiveBackground and Problemsmain workRelated workMethod(s)ConclusionReference(optional)Arouse for meOn Learning Associations of Faces and Voicescode: https://github.com/changil/f

文章目录On Learning Associations of Faces and VoicesSummaryResearch ObjectiveBackground and Problemsmain workRelated workMethod(s)experiment and anaysisConclusionReference(optional)Arouse for meOn Learning Associations of Faces and Voicesconference : SIGIR

Deep Audio-Visual learning: A SurveySummary写完笔记之后最后填，概述文章的内容，以后查阅笔记的时候先看这一段。注：写文章summary切记需要通过自己的思考，用自己的语言描述。忌讳直接Ctrl + c原文。Research Objective作者的研究目标。Problem Statement问题陈述，需要解决的问题是什么？Method(s)作者解决问题的方法/算法是什么？是否基于前人的方法？Evaluation作者如何评估自己的方法，实验的set

Title文章标题Summary写完笔记之后最后填，概述文章的内容，以后查阅笔记的时候先看这一段。注：写文章summary切记需要通过自己的思考，用自己的语言描述。忌讳直接Ctrl + c原文。Research Objective作者的研究目标。Problem Statement问题陈述，需要解决的问题是什么？Method(s)作者解决问题的方法/算法是什么？是否基于前人的方法？Evaluation作者如何评估自己的方法，实验的setup是什么样的，有没有问题或者可以借鉴的地方。Con

文章目录Face-Voice Matching using Cross-modal EmbeddingsSummary摘要 (中文)Research ObjectiveBackground and Problemsmain workRelated workMethod(s)ConclusionReference(optional)Arouse for meFace-Voice Matching using Cross-modal EmbeddingsSummaryThe author using o

文章目录Deep Residual Shrinkage Networks for Fault DiagnosisSummary摘要 (中文)Research ObjectiveBackground and Problemsmain workRelated workMethod(s)ExperimentConclusionReference(optional)Arouse for me写作流程Code reviewDeep Residual Shrinkage Networks for Fault D

OKR1.Getting some classial research in your research area.2. Mater some simple innvoation and difficult innvoation in research area.3. Master a clear route whcih can berak simple innvoation.1.Getting some classial research in your research area.综述

OMaster base self-researchKR1.whole big research’s classial work.2.Uderstanding what is yourself research’s area in whole big resarch.3. Some additional basic knowledge in your self research.1. whole research’s classial work.深度学习综述综述残差网络残差网络3

OHow to start and maintain research effectivelyKRHow to start research.Maintain self-research

Omaster designe toolsKRword or latexpptmardown or notebookvision1. word or latex2.ppt3. notebook or markdown4. vision

OResearch: Master pipeline research tools and article classificationKRDocument management：zotero

OSkill: python,data vision and anaysis, deep learning,machine learning.KRpythondata vison and anaysismachine learningdeep learningpython

#OGetting academic research ability of PH.DKR1. Master research tools ,base knowlege .Research: Master pipeline research tools and article classification2.Skill: python,data vision and anaysis, deep learning,machine learning.2.How to start and ma

OMaster other potential research or previous research fieldKR1.Master other potential research2. previous research field1.Master other potential research多模态潜在的十个研究点2. previous research field基于边缘计算的视觉感知研究...

OHow to write a research paperKRHow to writing paper.writing template.TIP:Here is a good video about how to write a good research paper. Here

OHow to train a networkKR数据预处理模型训练只是毕竟一个函数能逼近的极限，数据清洗和特征才是决定识别的上限。数据的shuffle 和augmentation。这个没啥好说的, aug也不是瞎加，比如行人识别一般就不会加上下翻转的，因为不会碰到头朝下的异型种数据预处理方法一般也就采用数据归一化即可。模型选择如果数据很少，低于百万，直接拿imagenet预训练的模型微调即可。模型方面，可以先用2或3层LSTM试一下，通常效果都不错。结构化数据用机器学

How to writing paper.1. tutorial1. videoHere is a good video about how to write a good research paper. Here.There have two key results:it can read easy by people without area knowledge.main work and creatiy can be understand and judgement by editor

文章目录前言keras学习keras 建模keras的三个层级sequential model 的要点keras layer的理解sequential 简单演示Function API总结实战：使用function API复现2015年cvpr论文任务概述前言keras学习keras 建模目标：使用tf.keras建立深度学习的模型，对keras.layer有系统的理解。要求：keras的三个层级建议不要用sequential,直接用functional。序列属于傻瓜开发，会让你越来越不会

原文链接：添加链接描述

VG5HH-D6E04-0889Y-QXZET-QGUC8CY55A-F6G80-H85HQ-WNN5X-W38W4AC11H-4HZ05-08EDQ-APQGX-YCUC8ZG780-8EZ9M-M89LY-M5PEG-W2AZ8ZF3NU-D6XEJ-48E7Q-27YNC-PC8EDAV5M0-F4W8M-088PZ-36ZNV-NFHE2AU15R-A0Z4N-M88HY-HQ...

损失函数一直在下降，为什么识别率上不去。1.最常见的原因：过拟合过拟合值得单独开个章节。主要包括1.数据量小，网络复杂2.learning rate 比较高，又没有设置任何防止过拟合的机制解决方法主要包括1.简化模型，利用现有深度学习手段增加数据（翻转，平移，随机裁剪，imgaug）2.利用 dropout层3.利用正则化2.你犯了错误：没有把数据规格化图片的话，img/255...

前言浏览网站时偶尔会看到一些十分好用的工具，这里专门开一个博客去记录用法。深度学习模型架构图转换这个github项目很贴心，它可以将现在流行框架的.H5模型文件自动转换成网络架构。这样有什么好处呢？首先就是便于计算每一层的输入和参数，看别人的论文和展示自己的成果时候都挺有用。使用步骤：打开链接https://lutzroeder.github.io/netron/** 将模型文件拖入网页 **得到如下的结果：特征分布可视化工具...

文章目录前言机器学习的基本数学知识微积分导数泰勒展开式偏导数线性代数向量矩阵概率论总结最优化方法最优化基本概念为什么要用迭代法梯度下降法牛顿法优化算法的问题凸优化问题拉格朗日对偶KKT条件本节总结前言参照https://www.bilibili.com/video/BV1Mb411c74N?p=1学习机器学习的数学理论知识。作者承诺这门课完全跟下来可以对以后学术研究和工程部署打下坚实的基础。拭目以待吧。课程内容：机器学习基本概念数学知识：微积分，线性代数，概率论和最优化方法。常用概念：过拟合，方

前言阅读了不少专题参考文献和综述后，自己对多模态的理解进一步加深。同时我针对多模态中的音视频融合也有了更深的理解。在此我做个多模态融合之音视频融合总结，概述我阅读的几个文献的核心内容，同时针对一些没有阅读到的，但比较有参考价值和重大意义的文献进行引用。我希望通过这篇文章总结能开拓我以后做研究的视野和方向，希望投稿论文别再出现如对该领域理解不够深刻的修改意见。文章主要内容来源：Baltrušaitis T, Ahuja C, Morency L P. Multimodal machine learni

文章目录1 学习tensorflow2.0之前需要的基础知识前言学习tf2.0需要基础知识python基础装饰器迭代器上下文管理器面向对象编程思想python的模块导入机制pip使用和virtualenvnumpy编程多维切片索引view&©清楚理解numpy ndarray的轴ndarray在内存上的结构有基本的认识linux操作系统SSHLinux系统的文件系统linux系统的账户管理linux系统的驱动安装git和github机器学习知道统计机器学习为什么叫 统计么？统计机器学习处理的结

前言正文总结

音频https://zhuanlan.zhihu.com/p/34989414视频