重点 (Top highlight)
信息性 (Informational)
“Wisdom is not a product of schooling but of the lifelong attempt to acquire it.”
“智慧不是学校教育的产物，而是终身学习的产物。”
— Albert Einstein
- 艾尔伯特爱因斯坦
介绍 (Introduction)
The ability to understand information produced by the individuals at the cutting edge of research within Artificial Intelligence and the Machine learning domain is a skill that every serious machine learning practitioner should acquire.
在人工智能和机器学习领域中，了解个人在前沿研究中产生的信息的能力是每位认真的机器学习从业者都应该获得的技能。
To stay relevant and increase your knowledge, machine learning practitioners need to have an academic mindset and habit. AI, ML and DL are evolving at a fast pace, and we have to equip ourselves with the knowledge to keep up with the field, knowledge that is only attainable through research papers.
为了保持相关性并增加您的知识，机器学习从业人员需要具有学术思维和习惯。 AI，ML和DL的发展日新月异，我们必须掌握与该领域保持一致的知识，而这些知识只能通过研究论文才能获得。
This article will provide you within instructions on how to go through a research paper effectively, and also provide the following:
本文将为您提供有关如何有效研究论文的指导，并提供以下内容：
- A systematic approach to reading a collection of papers to gain knowledge within a domain 一种系统的阅读论文集的方法，以获取领域内的知识
- How to properly read a research paper 如何正确阅读研究论文
- Useful online resources that can aid you in searching for papers and key information 有用的在线资源，可以帮助您搜索论文和关键信息
For those who would like to get to the key content within this article, scroll down to the section titled “Reading Research Papers”.
对于那些想要了解本文中关键内容的人，请向下滚动至标题为“ 阅读研究论文 ”的部分。
首先，吴安达是谁？ (First, Who is Andrew Ng?)
The information I provide in this article was derived from a Stanford lecture taught by Andrew Ng. I’ve also supplemented the information contained in this article with personal tips and information from resources on the internet.
我在本文中提供的信息来自Andrew Ng教授的斯坦福大学演讲 。 我还用个人提示和来自Internet资源的信息对本文中包含的信息进行了补充。
But first, a brief introduction on Andrew Ng.
但首先，简要介绍一下吴恩达 。
Andrew Ng is probably the most known(and watched) machine learning teacher on the internet. He is also the co-founder of both Deeplearning.ai and Coursera.
吴恩达 ( Andrew Ng )可能是互联网上最著名的( 也是观看最多的 )机器学习老师。 他还是Deeplearning.ai和Coursera的共同创始人。
Apart from his ongoing work within online education, he’s also a professor at Stanford University.
除了从事在线教育工作外，他还是斯坦福大学的教授。
More information on Andrew Ng is just a google search away.
有关Ng Andrew的更多信息，仅需Google搜索即可 。
It’s natural for a person to pick up skills and habits demonstrated by individuals around their environment; this is why most PhD students will acquire the skill of effectively digesting the content of a research paper appropriately. This is somewhat a fact, and Andrew mentions it very early in the video referenced earlier.
一个人自然会领会周围环境所展现出的技能和习惯。 这就是为什么大多数博士生将获得有效地适当消化研究论文内容的技能。 这是事实，Andrew在前面提到的视频中很早就提到了这一点。
But we are not PhD students, well some might be, but how do we normal individuals gain the required skills to read a research paper and understand its content wholeheartedly.
但是我们不是博士学位生，也许不是，但我们正常人如何获得阅读研究论文并全心全意理解其内容所需的技能。
阅读研究论文 (Reading Research Papers)
Specialization within the machine learning domain is favourable if you are talented. For example, having a generalist knowledge on the field of Computer vision is commendable, but having specialized knowledge and expertise within key techniques such as Pose estimation will be more appealing to companies and organization looking for practitioners within that domain.
如果您有才华，那么在机器学习领域内的专业化将非常有用。 例如，具有计算机视觉领域的通才知识是值得称赞的，但是在关键技术(例如Pose估计)中拥有专门知识和专业知识将对在该领域寻找从业人员的公司和组织更具吸引力。
So let’s use Pose Estimation as a guide to how we would approach reading research papers related to the subject matter: pose estimation.
因此，让我们以“姿势估计”为指南，介绍如何处理与主题相关的研究论文：姿势估计。
1.集中关注主题的资源集合。 资源可以是研究论文，中型文章，博客文章，视频，GitHub存储库等形式。 (1. Assemble collections of resources that focus on the subject matter. Resources can come in the form of research papers, Medium articles, blog posts, videos, GitHub repository etc.)
A quick google search on the phrase “pose estimation” will provide you with top resources that contain information in regards to the subject matter. At this initial step, the aim is to collate all resources that are relevant, such as YouTube videos, implementation documentations and of course research papers. Ideally, at this stage, there is no limit to the number of resources you consider important, but be sure to create a shortlist of papers, videos and articles that are useful.
谷歌对“姿势估计”一词的快速搜索将为您提供热门资源，其中包含有关主题的信息。 在此初始步骤中，目标是整理所有相关资源，例如YouTube视频，实施文档以及研究论文。 理想情况下，在此阶段，对您认为重要的资源数量没有限制，但是请确保创建有用的论文，视频和文章的清单。
2.在下一步中，您将深入研究与主题相关的任何资源。 至关重要的是，有一种方法可以跟踪对每个入围资源的理解。 吴安德(Andrew Ng)建议根据您的理解水平绘制的资源表与下表类似。 (2. In this next step, you will conduct a deep dive of any resource you deem relevant to the subject matter. It is crucial that there’s a method to track the understanding of each shortlisted resources. Andrew Ng, suggests a table of resource plotted against your understanding level that looks similar to the table below.)
It is advisable to ensure you go through at least 10–20% of the content of each paper you have added to the list; this will ensure that you have been exposed to enough of the introductory content within an identified resource and are able to gauge its relevancy accurately.
建议确保您浏览添加到列表中的每篇论文的至少10％到20％的内容； 这样可以确保您在已标识的资源中可以接触到足够的介绍性内容，并能够准确地评估其相关性。
For the more relevant papers/resources identified, it is expected that you progress to a higher level of understanding. Eventually, you will have identified some appropriate resources with content that you understand fully.
对于已确定的更相关的论文/资源，可以期望您获得更高的理解水平。 最终，您将确定一些适当的资源，这些资源具有您完全理解的内容。
You are probably asking yourself, “what number of papers/resource is sufficient”.
您可能会问自己：“多少论文/资源就足够了”。
Well, I don’t have the answer, but Andrew does.
好吧，我没有答案，但是安德鲁有答案。
According to Andrew, an understanding of 5–20 papers will showcase a basic understanding within the subject matter, perhaps enough understanding to progress to implementation of techniques.
据安德鲁说，对5到20篇论文的理解将展示该主题内的基本理解，也许足以理解技术的实施。
50–100 papers will primarily provide you with a very good understanding of the domain.
50-100篇论文将主要为您提供对该领域的很好的理解。
After going through the resources and extraction of vital information, your table might look something similar to what’s shown below.
在浏览完资源并提取了重要信息之后，您的表可能看起来类似于以下所示。
3.第三步是我试图理解研究论文时所观察到的对我有用的快速提示。 第三步是进行结构化注释，以您自己的话语总结论文中的关键发现，发现和技术。 (3. The third step is a quick tip from what I’ve observed works for me when trying to understand research papers. The third step is to take structured notes that summarises the key discoveries, findings and techniques within a paper, in your own words.)
The following steps will now be focused on how to read a single research paper.
现在，以下步骤将集中于如何阅读一份研究论文。
阅读一份研究论文 (Reading A Single Research Paper)
Reading for the purpose of understanding is not done through one pass of the contents within the paper. According to Andrew, reading a paper from the first word to the last word in one sitting might not be the best way to form an understanding.
出于理解目的而进行的阅读不是通过一篇论文中的内容来完成的。 据安德鲁说，一次坐着从头到尾阅读论文可能不是形成理解的最佳方法。
Be prepared to go through a paper at least three times to have a good understanding of its content
准备至少阅读三次论文以充分理解其内容
4.在第一步中，首先阅读本文中的以下部分：标题，摘要和图形。 (4. In your first pass, start with reading the following sections within the paper: title, abstract and figures.)
5.第二遍需要您阅读以下各节：简介，结论，另一遍图形以及内容的其余部分。 (5. The second pass entails you reading the following sections: introduction, conclusion, another pass through figures and scan through the rest of the content.)
The introduction and conclusion section of a paper contains clear and concise information on the content of the paper and a summary of any findings. The information presented in this section usually dismisses any supplementary information and only key information are included. This is beneficial to you as a reader as you get the vital information required to proceed to the other sections within the paper.
论文的引言和结论部分包含有关论文内容的简洁明了的信息以及所有发现的摘要。 本节中提供的信息通常忽略任何补充信息，并且仅包含关键信息。 作为读者，这对您有好处，因为您获得了继续阅读本文其他部分所需的重要信息。
6.论文的第三阶段通过阅读论文的整个部分，但跳过了可能对您不熟悉的任何复杂的数学或技术公式。 在此过程中，您还可以跳过任何您不理解或不熟悉的术语和术语。 (6. The third pass of the paper involves reading the whole sections within the paper but skipping any complicated maths or technique formulations that might be alien to you. During this pass, you can also skip any terms and terminologies that you do not understand or aren’t familiar.)
7.对某个领域进行深入研究的人员可能还需要通过一些考试。 这些附加的通行证将主要集中于对本文中介绍的数学，技术和未知术语的理解。 (7. Those conducting in-depth research into a domain can take a few more passes. These additional passes will mainly be focused on an understanding of the maths, techniques and unknown terminologies presented within the paper.)
For those who are generally reading research papers for informational and engineering purposes, then in-depth research might be very time consuming, especially if you have 20 more papers to get through.
对于那些通常出于信息和工程目的阅读研究论文的人，那么深入研究可能会非常耗时，尤其是如果您还有20篇论文需要阅读。
I went through the process presented in this article with the original paper introducing the LeNet convolutional neural network, and I summarised the key content in notes which I later then converted to a series of Medium articles.
我通过介绍LeNet卷积神经网络的原始论文来完成本文介绍的过程，并总结了笔记中的关键内容，随后将其转换为一系列中型文章。
问自己的问题 (Questions To Ask Yourself)
Andrew provides a set of questions that you should ask yourself as you read a paper. These questions generally will show you understand the critical information presented in a paper. I use the questions below as beacons to ensure I don’t stray from the aim of understanding vital information.
安德鲁提供了一系列问题，您在阅读论文时应该问自己。 这些问题通常会告诉您了解论文中提供的关键信息。 我将以下问题用作信标，以确保我不会偏离理解重要信息的目的。
They are as follow:
它们如下：
- Describe what the authors of the paper aim to accomplish, or perhaps did achieve. 描述本文的作者旨在实现或可能实现的目标。
- If a new approach/technique/method was introduced in a paper, what are the key elements of the newly proposed approach? 如果在论文中引入了新的方法/技术/方法，那么新提议的方法的关键要素是什么？
- What content within the paper is useful to you? 论文中哪些内容对您有用？
- What other references do you want to follow? 您还要遵循哪些其他参考？
辅助研究的其他资源 (Additional Resources To Assist Research)
Several online resources have made the discovery and retrieval of relevant information relatively easy. Below are examples of resources that will assist you in your search for pertinent information.
几个在线资源使相关信息的发现和检索相对容易。 以下是可帮助您搜索相关信息的资源示例。
结论 (Conclusion)
“Learn steadily rather than short burst for longevity.“
“要稳定学习，而不要短暂学习才能长寿。”
— Andrew Ng
－吴彦祖
I’m still relatively new to the field of Machine Learning and Computer Vision, there is a lot that I do not know (that’s an understatement). Still, I believe that if an individual is consistent in their search for knowledge, regardless of the domain, they’ll be rewarded with an understanding and expertise that surpasses the norm.
我对机器学习和计算机视觉领域还比较陌生，有很多我不知道的地方( 这是一种轻描淡写 )。 尽管如此，我仍然相信，如果一个人在寻求知识方面始终如一，无论其领域如何，他们都会获得超越标准的理解和专业知识。
From the techniques introduced by Andrew Ng, I’ll be reading at least four research papers a month, reading to the point of understanding. I’ll be honest and say that the LeNet paper took me about a week and a half to complete wholeheartedly. But you get better and faster at reading and understanding research papers the more times you do it.
根据吴安德(Andrew Ng)提出的技术，我每个月至少阅读四篇研究论文，以期达到理解的目的。 老实说，LeNet论文花了我一个半星期的时间才全心全意地完成。 但是，这样做的次数越多，您在阅读和理解研究论文上就会越快越好。
Andrew states in his video that he carries a batch of research papers around with him, intending to read them. Andrew is a prominent figure within the field of machine learning, and I believe emulating his habits and learning techniques can be advantageous to your learning journey.
安德鲁在影片中表示，他随身携带了一批研究论文，打算阅读这些论文。 安德鲁是机器学习领域的杰出人物，我相信效法他的习惯和学习技巧可能对您的学习之旅有利。
希望您觉得这篇文章有用。 (I hope you found the article useful.)
To connect with me or find more content similar to this article, do the following:
要与我联系或查找更多类似于本文的内容，请执行以下操作：
2069
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
