我想学习人工智能和机器学习,我可以从哪里开始呢(四)
I want to learn artificial intelligence and machine learning. Where can I start?译文简介
网友:请看,人工智能(AI)是当今最受欢迎的技能之一,是一个涵盖性术语,包括ML(机器学习)和DL(深度学习)。DL又是ML的一个子集,因此,如果一个人需要掌握AI,首先需要磨练他/她的ML和DL的技能。而且,当涉及到自主学习机器学习时,决定一个有效的资源是至关重要的......
正文翻译
I want to learn artificial intelligence and machine learning. Where can I start?
我想学习人工智能和机器学习,我可以从哪里开始呢?
评论翻译
很赞 ( 0 )
收藏
See, AI(Artificial Intelligence) - one of the most sought after skills in today’s time, is an umbrella term and encompasses ML(Machine Learning) and DL(Deep learning). DL again is a subset of ML. Hence, if one needs to master AI, one would first need to hone his/her skills in ML and DL.
And, when it comes to self-learning Machine Learning, it is essential to decide on an effective resource - the one that considers that the students are new to the domain and are not adept with the Machine Learning environment, the one that explains why the program is executing the way it is executing, the one that discusses different approaches the students can use to solve a question and does not restrain their mind to just one method, the one that does not merely glide over the topics. Well, I too learned these lessons the hard way.
But now, having mastered Machine Learning and thence having bagged a high-paying Machine Learning Development job fresh out of college at Airbnb, after facing multiple challenges, I believe, I should put an answer to this question so as to make your learning less troublesome and less time consuming than mine.
请看,人工智能(AI)是当今最受欢迎的技能之一,是一个涵盖性术语,包括ML(机器学习)和DL(深度学习)。DL又是ML的一个子集,因此,如果一个人需要掌握AI,首先需要磨练他/她的ML和DL的技能。
而且,当涉及到自主学习机器学习时,决定一个有效的资源是至关重要的——一个考虑到学生是新的领域和不熟练的机器学习环境,一个解释为什么程序执行的方式,它讨论了学生可以用来解决一个问题的不同方法,而不是把他们的思想限制在一种方法上,它不只是跳过主题。好吧,我也经历了惨痛的教训。
但现在,我已经掌握了机器学习,大学刚毕业就在爱彼迎找到了一份高薪的机器学习开发工作,在经历了多次挑战之后,我相信我应该回答这个问题,让你的学习遇到的麻烦没我多,花费的时间也没我多。
Moreover, in several resources I found that while solving questions, the author applies a logic/technique that has not been taught to the learner yet. This leads the learner to skip to those sections of the tutorial, where that particular topic is discussed. The concepts taught in those sections in turn apply logic that belongs to another concept. Often, this is a repetitive cycle.
See, as a beginner, much of the learner’s interest in the subject lies in the hands of the tutor and the manner in which the course is delivered. Inefficient coaching can pretty quickly lead to the learner losing interest in the subject, and in worst cases - this can significantly hamper one’s career.
坦率地说,为了掌握机器学习,我想尽了一切办法。我买了太多的课程、书籍和PDF材料,但我总是在学习的几天就碰壁了。在大多数情况下,我觉得作者/导师急于结束课程,没有教育编写这些代码背后的基本原理,并假设自己精通机器学习环境。但对于初学者来说,情况并非如此。
此外,在一些资源中,我发现在解决问题时,作者应用了一种尚未教给学习者的逻辑/技术。这导致学习者跳到教程中讨论特定主题的那些部分。这些章节中教授的概念依次应用属于另一个概念的逻辑。通常,这是一个重复的循环。
作为一名初学者,学习者对这门学科的兴趣很大程度上取决于导师和讲授课程的方式。低效的指导会很快导致学习者对该学科失去兴趣,在最坏的情况下,这会严重阻碍一个人的职业生涯。
原创翻译:龙腾网 https://www.ltaaa.cn 转载请注明出处
The artificial intelligence (AI) and machine learning (ML) industry is growing by leaps and bounds. The global market size for both these sectors was about $9.51 billion in 2018 and it is expected to rise to $118.6 billion by 2025.
In such a booming environment, if you want to learn AI and ML, you will have to look beyond your immediate obxtives. To begin with, find out the following:
What Is Your Goal?
The first question you need to answer is what precisely your goal is. Your learning process will drastically vary with your answer.
For instance, if you intend to learn AI and ML for getting a job, you'll have to focus on obtaining practical experience. However, if you want to learn them as a part of academics, you'll need to concentrate more on the theoretical aspects.
What's Your Knowledge Level?
Secondly, you will have to figure out how much you already know about AI and ML. Self-learning can work if you possess a basic understanding of:
人工智能(AI)和机器学习(ML)行业正在飞速发展。这两个行业的全球市场规模在2018年约为95.1亿美元,预计到2025年将增至1186亿美元。
在这样一个蓬勃发展的环境中,如果你想学习AI和ML,你必须超越眼前的目标。首先,找出以下内容:
你的目标是什么?
你需要回答的第一个问题是你的目标到底是什么。你的学习过程会因你的答案而大不相同。
例如,如果你想学习AI和ML以获得工作,你必须专注于获得实践经验。然而,如果你想把它们作为学术的一部分来学习,你需要更多地关注理论方面。
你的知识水平如何?
其次,你必须弄清楚你对人工智能和机器学习已经了解了多少。如果你对以下方面有基本的了解,那么自我学习是可行的:
Markup languages.
Foundational mathematics.
Numerical computation.
Multi-variable calculus.
Linear algebra.
Probability theory.
Do You Have Time?
Last but not the least, ascertain the exact amount of time that you have in hand. Irrespective of your goals, knowledge, or time, upGrad can be a great place to start learning about AI and ML. With its specifically designed programs, one-on-one mentorship, and extensive knowledge base, you can learn AI and ML in as few as 11 months!
编程:尤其是Python。
标记语言。
基础数学。
数值计算。
多变量微积分。
线性代数。
概率论。
你有时间吗?
最后但并非最不重要的是,确定你手头的确切时间。无论你的目标、知识或时间如何,upGrad平台都是开始学习AI和ML的好地方。凭借其专门设计的课程、一对一的指导和广泛的知识库,你可以在短短11个月内学习AI与ML!
How do I start learning machine learning if I know programming?
Build something; understand the theory behind what you’ve built; repeat.
A lot of people suggest picking up a book like Elements of Statistical Learning and just cranking through that. That might work and is probably the most efficient way of picking up the basics, but it’s kind of a boring path.
Instead, pick a dataset that you’re interested in. Don’t spend time collecting data: it just has to be something that’s easily available, e.g. elections, sports, etc. Identify a prediction problem related to that dataset and try to build a predictive model using libraries like scikit-learn, statsmodels, etc. or their equivalents in R.
Your models’ predictions will probably be pretty bad at first and you’ll find yourself just cycling through different models in order to get better metrics. Resist this temptation. Instead, just pick up a few simple models (e.g. a linear model, maybe an SVM) and then read up on the relevant details from a textbook (e.g. Elements of Statistical Learning or Pattern Recognition and Machine Learning). Of course, if you don’t have the math or CS background needed to understand a particular concept, you should take the time to learn that as well.
Once you feel like you’ve developed a reasonably deep understanding of what you’re doing, attack the same or a different problem again.
If you learn things this way, it’s much harder to forget the details since you’ll end up using a lot of them immediately. Secondly, the entire process will be much more pleasant and interesting than trying to force-feed yourself an entire book with no clear goal in sight.
如果我 懂编程,我如何开始学习机器学习?
构建一些东西;理解你所构建的理论,再重复
很多人建议读一本《统计学习要素》之类的书,然后仔细阅读。这可能会起作用,可能是最有效的学习基础知识的方法,但这是一条枯燥的道路。
相反,选择你感兴趣的数据集。不要花时间收集数据:它必须是容易获得的东西,例如选举、体育等。确定与该数据集相关的预测问题,并尝试使用scikit learn、statsmodels等库或R中的等效库构建预测模型。
你的模型的预测起初可能会非常糟糕,你会发现自己只是在不同的模型中循环,以获得更好的指标。抵制这种诱惑。相反,只需选择一些简单的模型(例如线性模型,可能是SVM),然后阅读教科书中的相关细节(例如统计学习的元素或模式识别和机器学习)。当然,如果你没有理解一个特定概念所需的数学或计算机科学背景,你也应该花时间去学习。
一旦你觉得自己对自己正在做的事情有了相当深刻的理解,就可以再次解决相同或不同的问题。
如果你用这种方式学习,那么忘记细节就更难了,因为你最终会立即使用很多细节。其次,整个过程将比试图在看不到明确目标的情况下强迫自己吃透一整本书更愉快和有趣。
How can I start programming machine learning and artificial intelligence?
I’m in no way qualified to fully answer this question - I just asked a very similar question myself a few days ago.
However, taking into account that you are just starting your CS degree, I might have tip or two for you. While what I’m about to say is currently related to AI/ML, you can consider it general advice.
Have you ever experienced what I call “maths enlightenment”? When a concept that felt totally useless finally made sense? I’ve been there several times before: every now and then you realize that after all, derivatives, integrals or differential calculus might not be total bullshit. Or, taking a step back, do you remember when you first realized in practice why it made sense to get yourself to understand multiplication? That you finally knew how many ice creams you can buy on your pocket money?
Machine learning is the first area where after a very long while I felt like I should’ve paid more attention when learning about matrices, differential algebra or any of the “hard” courses, in general. While some of the most practical courses might seem more appealing (I remember being super excited about “iOS development”), and certainly they can be more useful in the very short run, the hard subjects (physics, algorithms, calculus, signals and systems) will pay off forever.
I am saying this as a CS dropout: while you might feel well-prepared to use the technologies and the tools available today, every now and then a new set of technologies pop up where you will sort of be required to know about the low-level stuff in order to understand the bigger picture. That’s where I am today: learning some advanced math that I previously neglected again, finally being able to appreciate some concepts .
我如何开始编写机器学习和人工智能程序?
我没有资格完全回答这个问题——几天前我自己也问了一个非常类似的问题。
然而,考虑到你刚刚开始你的计算机科学学位,我可能有一两个建议给你。虽然我要说的是目前与人工智能和机器学习相关的内容,但你可以将其视为一般性建议。
你有没有经历过我所说的“数学启蒙”?当一个感觉完全无用的概念终于有了意义?我以前经历过好几次:时不时你会意识到,毕竟,导数、积分或微分学可能都不是废话。或者,退一步,你还记得当你在实践中第一次意识到为什么理解乘法是有意义的吗?你终于知道你可以用零花钱买多少冰淇淋了?
机器学习是第一个领域,在很长一段时间后,我觉得在学习矩阵、微分代数或任何“核心”课程时,我应该多加注意。虽然一些最实用的课程看起来更吸引人(我记得我对“iOS开发”非常兴奋),而且在很短的时间内肯定会更有用,但难学的科目(物理、算法、微积分、信号和系统)将永远有回报。
我说这句话是因为一个计算机科学方面辍学者:虽然你可能已经准备好使用当今可用的技术和工具,每隔一段时间,就会出现一套新的技术,要求你了解底层的东西,了解事物的全貌。这就是我现在的处境:重新学习一些以前被我忽视的高等数学,终于能够欣赏一些概念。
These last several years have seen remarkable growth for AI. Already, artificial intelligence (AI) and Machine Learning is producing billions of dollars in revenue across a variety of businesses, as well as offering up a plethora of job opportunities.
This trend in artificial intelligence will continue, as the majority of industry verticals embrace the promise of AI to create better tomorrow while also opening up a multitude of employment opportunities. These cutting-edge technologies are available to future professionals, allowing them to establish a long, rewarding, and satisfying career.
Similarly, for true machine learning to work, the computer must be able to identify patterns without being explicitly taught how to do so. Using artificial intelligence, robots may learn a task via experience without being specifically programmed for it.
过去几年,人工智能取得了显著的增长。人工智能(AI)和机器学习已经在各种业务中创造了数十亿美元的收入,并提供了大量的就业机会。
人工智能的这一趋势将继续下去,因为大多数垂直行业都信奉人工智能的承诺:创造更好的明天,同时也创造了大量的就业机会。这些尖端技术可供未来的专业人士使用,使他们能够建立一个长期的、有回报的、令人满意的职业生涯。
类似地,要想让真正的机器学习起作用,计算机必须能够识别模式,而不需要明确地教授如何这样做。使用人工智能,机器人可以通过经验学习任务,而无需专门为其编程。
An approximate road map to becoming proficient in Machine Learning and Artificial Intelligence may be found here. So hold on as I guide you.
Understand the pre-requisites - To begin, you must understand the prerequisites. Most people need to first master Linear Algebra, Multivariate Calculus, Statistics and Python before they can dive into ML and AI like a prodigy. A basic understanding of these topics is sufficient to get started, although a PhD in these fields is not required. For Machine Learning, Linear Algebra, as well as Multivariate Calculus, are both necessary skills. However, the extent to which you need them depends on your role as a data scientist. Around 80% of your work as an ML professional will be spent acquiring and cleaning data. Data collection, analysis, and presentation are all aspects of statistics. This means you'll need to study it, which shouldn't be a surprise to you. There's also one thing you can't ignore - Python.
理解机器学习和人工智能并不是一件容易的事情。首先向计算机提供一组高质量的数据,然后使用基于数据的各种机器学习模型和已经开发的许多方法对计算机进行训练。因此,你使用的算法将取决于你正在自动化的数据类型和任务类型。
这里可以找到精通机器学习和人工智能的大致路线图。所以,在我引导你的时候,坚持住。
了解先决条件-首先,你必须了解先决要求。大多数人需要先掌握线性代数、多元微积分、统计学和Python语言,然后才能像神童一样潜入人工智能和机器学习。尽管不需要这些领域的博士学位,但对这些主题做到基本了解就足以开始了。对于机器学习,线性代数和多元微积分都是必要的技能。然而,你需要它们的程度取决于你作为数据科学家的角色。作为机器学习专业人员,你大约80%的工作将用于获取和清理数据。数据收集、分析和呈现都是统计的各个方面。这意味着你需要研究它,这对你来说并不奇怪。还有一件事你不能忽视——Python语言。
Take part in projects - The fascinating and entertaining part of Machine Learning comes after understanding the foundations. PROJECTS! Through the integration of your largely theoretical understanding with real-world application, you will be able to improve your machine learning skills. To get you started, the 'Titanic: Machine Learning from Disaster project,' the 'Digit Recognizer project,' etc. can help you gain confidence. It is also possible to use the projects as a tool while looking for employment in this sector.
A smart option when you've learned the basics is to network with other data scientists or enrol in professional training courses. Furthermore, courses like as upGrad, Skillslash, Edureka are some of the top AI and ML platforms on the market right now. There are a number of reasons why Skillslash, for example, provides some of the best intermediate-level AI and ML courses.
熟悉不同的机器语言和人工智能思想——模型、特征(标签)、目标(标签),训练和预测是其中几个基本概念。机器学习包括有监督、无监督、半监督和强化学习。为了在机器学习中使用数据,需要花费大量精力来收集、整合、清理和准备数据。你需要高质量的数据,但随机生成大量数据并不少见。其他要求包括研究替代模型和使用真实世界数据。这可以帮助你了解在特定情况下什么样的模型是合适的。还有学习使用各种模型分析数据的正确方法的问题。了解各种型号上使用的各种调整设置和调整方法可以使这更容易实现。
参与项目-机器学习的迷人和有趣的部分是在了解基础之后参与项目!通过将你的基本理论理解与实际应用相结合,你将能够提高你的机器学习技能。为了让你开始,“泰坦尼克号:灾难中的机器学习项目”、“数字识别器项目”等可以帮助你获得信心。你也可以将这些项目作为在该行业寻找就业机会的工具。
当你学会了基础知识后,一个明智的选择是与其他数据科学家建立网络或参加专业培训课程。此外,upGrad、Skillslash、Edureka等课程是目前市场上一些顶级的人工智能和机器学习平台。例如,Skillslash提供一些最好的中级的人工智能和机器学习课程是有很多原因的。
Individually tailored courses are available through Skillslash for a moderate fee.
Over the course of the nine-month term, students who sign up for Skillslash get access to 350+ hours of live sessions.
In addition, they give students the chance to collaborate on real-world projects companies. For students to enhance their portfolios and receive a project experience certificate, they can engage in joint projects with businesses.
Courses, such as Data Science and AI Program, Full-Stack AI and ML Program, are available to students from a range of professional backgrounds.
In order to keep expenses down, they've created affordable courses that costs Rs. 89,000 and Rs. 35,000 for professionals and newcomers respectively.
Final Thoughts - Self-driving cars, chatbots, realistic speech recognition, efficient internet search, and a vastly improved understanding of the human genome have all been made possible by artificial intelligence and machine learning in recent years. Most of us utilise AI and ML hundreds of times a day, without ever recognising it. That's why all learners receive intermediate-level tuition from Skillslash.
一位行业专家审查了他们的项目经验,这在该领域非常重要。
通过Skillslash提供个性化定制课程,收费适中。
在九个月的学期中,报名参加Skillslash的学生可以获得350多个小时的直播课程。
此外,他们还为学生提供了在现实世界项目公司中合作的机会。为了提高学生的投资组合并获得项目经验证书,他们可以与企业联合开展项目。
课程,如数据科学和人工智能计划、全栈人工智能和ML计划,可供各种专业背景的学生学习。
为了降低开支,他们为专业人士和新人开设了价格合理的课程,分别为89000卢比和35000卢比。
最后的想法-近年来,人工智能和机器学习使自动驾驶汽车、聊天机器人、逼真的语音识别、高效的互联网搜索以及对人类基因组的理解大大提高成为可能。我们中的大多数人每天使用AI和ML数百次,但从未意识到这一点。这就是为什么所有学习者都会从Skillslash接受中级水平的学费的原因。
One of the most common questions in today’s day and age is “How to start learning Artificial Intelligence and Machine Learning?”. This is not an easy question to answer. One needs to go through various online e Books, websites, and blogs, courses, classroom training, training institutes, and on-the-job training to learn this. This answer, however, leads to a range of other questions that need to be answered, such as which book to buy or which online course to take.
In order to begin, here is a list of details of the sources mentioned above:
E-books: Reading a book or two related to a field is one of the best and simplest ways to learn about anything. Some of the e-books available for data science can be read for a good start. Some of them can be downloaded for free in order to get a good understanding of data science. E-books offer the unique advantage of allowing a person to gain knowledge at their own pace without relying on other individuals. As a bonus, it's one of the best and most affordable ways to learn about Data Science, as well as in-depth. However, e-books are primarily hindered by a lack of support. The issue is accentuated if the person does not have any prior background in the field. Furthermore, it is difficult to learn from books due to a lack of practical experience and knowledge, which is a major drawback in a field such as this.
当今时代最常见的问题之一是“如何开始学习人工智能和机器学习?”。这不是一个容易回答的问题。人们需要通过各种在线电子书、网站、博客、课程、课堂培训、培训机构和在职培训来了解这一点。然而,这个答案引出了一系列其他需要回答的问题,比如买哪本书或参加哪门在线课程。
首先,这里列出了上述来源的详细信息:
电子书:阅读一两本与某一领域相关的书是了解任何事物的最好和最简单的方法之一。一些可用于数据科学的电子书可以作为一个良好的开端。其中一些可以免费下载,以便更好地了解数据科学。电子书具有独特的优势,可以让一个人以自己的速度获得知识,而不依赖其他人。作为奖励,它是学习数据科学以及深入学习的最好和最实惠的方法之一。然而,电子书主要受到缺乏支持的阻碍。如果此人没有该领域的任何背景,这个问题就会更加突出。此外,由于缺乏实践经验和知识,很难从书中学习,这是此类领域的一个主要缺陷。
原创翻译:龙腾网 https://www.ltaaa.cn 转载请注明出处
For data science: Python Data Science Handbook by Jake VanderPlas
For Machine Learning: Understanding Machine Learning by Shai Shalev-Shwartz and Shai Ben-David
For Artificial Intelligence or Deep Learning: Deep Learning by Ian Goodfellow
Websites and blogs: An ocean of data science-related websites and blogs exist today, which leaves many asking, Which is the best place to learn machine learning and AI? Learning can be found primarily through websites and blogs, which often offer useful practical knowledge. The most popular websites and blogs are - Kdnuggets, Kaggle, Data Camp, etc. Additionally, Reddit's and Google Newsblogs also provide information on Data Science, which is a key component to the dissemination of news on Data Science. Moreover, the comments given by users on these blogs help elaborate certain concepts. Similar to learning from books, they are not dynamic and do not offer individual support.
Online courses: Based on my 5+ years of experience in this field, enrolling yourself in an online training program is the best way to build your future in the field of Artificial Intelligence and Machine Learning. When you are guided by professionals and industry experts, your chances of success increase. Many amazing platforms exist that teach machine learning concepts, so you should definitely use them. This again raises questions such as – Which course should I take to learn AI? You can find many e-learning courses on the internet provided by excellent online course providers. A few of them are listed below:
你可以参考的书籍列表:
对于数据科学:Jake VanderPlas的Python数据科学手册
《机器学习:理解机器学习》作者:Shai Shalev Shwartz和Shai Ben David
人工智能或深度学习:伊恩·古德费罗的深度学习
网站和博客:如今存在着大量与数据科学相关的网站和博客,这让许多人不禁要问,哪一个地方是学习机器学习和人工智能的最佳场所?学习主要可以通过网站和博客找到,这些网站和博客通常提供有用的实用知识。最受欢迎的网站和博客是-Kdnuggets、Kaggle、Data Camp等。此外,Reddit和谷歌新闻博客还提供数据科学信息,这是数据科学新闻传播的关键组成部分。此外,用户在这些博客上的评论有助于阐述某些概念。与从书本上学习类似,它们不是动态的,也不提供个人支持。
在线课程:基于我在该领域5年以上的经验,注册在线培训计划是在人工智能和机器学习领域构建未来的最佳方式。当你在专业人士和行业专家的指导下,你成功的机会会增加。存在着许多教授机器学习概念的令人惊叹的平台,因此你绝对应该使用它们。这再次引发了一些问题,比如——我应该选择哪门课程来学习人工智能?你可以在互联网上找到许多优秀在线课程提供商提供的在线学习课程。以下列出了其中一些:
Stanford University – Machine Learning – The course is available on Coursera. This course is taught by Google Brain founder Andrew Ng. In this course, you can choose whether to take it for free or to pay in order to get a certificate that will aid you in your career in the future. You will learn about examples of AI-driven technologies from real life, like advanced mechanisms of web search and speech recognition. You will also learn how neural networks learn.
In addition to this, there is one specific site I would like to suggest, which is Skillslash. Definitely one of the most effective ways to learn about artificial intelligence, data science, and machine learning. When I came across this platform during my research, I was impressed by the features offered here.
与谷歌AI一起学习——它由谷歌推出,旨在向公众传达什么是人工智能以及它是如何工作的。尽管该资源仍处于起步阶段,但它已经提供了一个机器学习课程,其中包含了谷歌的TensorFlow库。它介绍了TensorFlow并解释了神经网络是如何设计的,从初学者到专家,每个人都会发现它很有用。
斯坦福大学-机器学习-该课程可在Coursera平台上可获得。本课程由Google Brain创始人吴恩达教授。在这门课程中,你可以选择是免费还是付费,以获得一份有助于你未来职业生涯的证书。逆将了解现实生活中人工智能驱动技术的例子,如网络搜索和语音识别的高级机制。你还将学习神经网络系统是如何学习的。
除此之外,我还想推荐一个特定的网站,那就是Skillslash。绝对是学习人工智能、数据科学和机器学习最有效的方法之一。当我在研究期间遇到这个平台时,我对这里提供的功能印象深刻。