TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Ask HN: What maths are critical to pursuing ML/AI?

628 点作者 chrisherd超过 7 年前
What maths must be understood to enable pursuit of either of the above fields? are there any seminal texts/courses/content which should be consumed before starting?

47 条评论

CuriouslyC超过 7 年前
You absolutely need a solid grounding in multi-variable calculus, linear algebra, probability theory and information theory. It will also be helpful to be well versed in graph theory.<p>In my opinion one of the best starting points is &quot;Information Theory, Inference and Learning Algorithms&quot; by David MacKaye. It&#x27;s a bit long in the tooth now, but it is still one of the most approachable and well written books in the field.<p>Another old book that stands up very well is &quot;Probability Theory: the Logic of Science&quot; by E. T. Jaynes.<p>&quot;Elements of Statistical Learning&quot; by Tibshirani is also good.<p>&quot;Bayesian Data Analysis&quot; by Andrew Gelman is another great read.<p>&quot;Deep Learning&quot; by Ian Goodfellow and Yoshua Bengio is useful for getting caught up with recent advances in that field.
评论 #15118390 未加载
评论 #15117191 未加载
评论 #15117252 未加载
评论 #15116942 未加载
评论 #15116943 未加载
评论 #15117171 未加载
评论 #15118041 未加载
评论 #15119587 未加载
评论 #15117680 未加载
评论 #15118812 未加载
评论 #15120314 未加载
评论 #15117333 未加载
评论 #15118921 未加载
评论 #15119871 未加载
评论 #15116835 未加载
leecarraher超过 7 年前
It will depend on the level you plan to engage in the ML&#x2F;AI space. If you just want a job in ML&#x2F;AI , you are in luck. Due to the growing assortment of available, mostly to fully automated, solutions like Datarobot, H2O, sckit-learn, keras(w&#x2F; tensorflow) the only math you will absolutely &#x27;need&#x27; is probably just Statistics. Regardless of what&#x27;s going on behind the scenes with whatever automatically tuned and selected algorithm your chosen solutions uses, you will still need some stats in the end to show the brass that &#x27;your&#x27; model works. the upside is that then you can spend time, learning feature extraction, data engineering, and the aforementioned toolkits, in particular what models they make available.<p>If you want to develop new techniques and algorithms, the the skies the limit, you&#x27;ll of course want Stats too though.
评论 #15117607 未加载
评论 #15117767 未加载
评论 #15118065 未加载
irchans超过 7 年前
Basic probability is very helpful: Expectation, Standard deviation, P(A and B) = P(A)*P(B) is A and B are independent, P(A or B) = P(A)+P(B) if A and B are mutually exclusive. Also, knowing algebra is very helpful.<p>In a way, you don&#x27;t really need to know much more because there is a lot of good software out there.<p>If you want to learn more math, learn Linear Regression, Logistic Regression, p-values, probability density functions, cumulative density function, the Central Limit Theorem, Gaussian Distributions, Exponential Distributions, Binomial Distribution, (maybe) Student-T distribution.<p>If you want to learn even more, first learn matrices (adding, multiplying, inverting, rank, span, matrix decomposition (SVD, and eigendecomposition are the most important)).<p>If you want to learn even more, it&#x27;s time to learn calculus. Integral calculus is needed for continuous probability distributions and information theory. Differential calculus is needed to understand back propagation.<p>There are a lot of other good suggestions written by the other commentators.
评论 #15117706 未加载
KirinDave超过 7 年前
If you care about actually reading the nournals, as I do, and you had a very poor math education (as mine was abysmally opposed to both math and science as enemies of religion) then here are things I&#x27;ve determined I need to know to read journals:<p>- Core statistics. You need to be familiar with how statisticians treat data, because it comes up a lot.<p>- Calculus. You do not need to be a wizard at working the numbers but you do need to understand how to describe the process of differentiation and integration over multiple variables comfortably.<p>- Linear algebra. It&#x27;s essentially the basis for everything, even more than statistics.<p>- Numerical nethods for computing. I constantly have to refer to references to understand why people make the choices they do.<p>- Theory of computation and the research clustered around it. Familiarity here helps a lot. Sometimes I even catch errors or am able to recognize improvements available. Also there is a lot of crossover, as one would expect. An example: everyone is remembering how good automatic differentiation is! And given that properly combined differentiable equations are also differentiable, AD let&#x27;s you optimize over your optimization process. It&#x27;s differentiable turtles all the way down.<p>My next big challenge is nonparametric statistics. Many researchers tell me that this is a very fruitful place to be and many methods there are increasingly making improvements in ML.
评论 #15119980 未加载
mindcrime超过 7 年前
It depends on how deep you want to go and what your goals are, but I&#x27;d say that CuriouslyC pretty much nailed it. Multi-variable calculus, linear algebra, and probability &#x2F; stats are definitely the core.<p>If you&#x27;re interested in finding more &quot;freely available online&quot; maths references, check out:<p><a href="http:&#x2F;&#x2F;people.math.gatech.edu&#x2F;~cain&#x2F;textbooks&#x2F;onlinebooks.html" rel="nofollow">http:&#x2F;&#x2F;people.math.gatech.edu&#x2F;~cain&#x2F;textbooks&#x2F;onlinebooks.ht...</a><p><a href="http:&#x2F;&#x2F;www.openculture.com&#x2F;free-math-textbooks" rel="nofollow">http:&#x2F;&#x2F;www.openculture.com&#x2F;free-math-textbooks</a><p><a href="https:&#x2F;&#x2F;open.umn.edu&#x2F;opentextbooks&#x2F;SearchResults.aspx?subjectAreaId=7" rel="nofollow">https:&#x2F;&#x2F;open.umn.edu&#x2F;opentextbooks&#x2F;SearchResults.aspx?subjec...</a><p><a href="https:&#x2F;&#x2F;ocw.mit.edu&#x2F;courses&#x2F;online-textbooks&#x2F;#mathematics" rel="nofollow">https:&#x2F;&#x2F;ocw.mit.edu&#x2F;courses&#x2F;online-textbooks&#x2F;#mathematics</a><p><a href="https:&#x2F;&#x2F;aimath.org&#x2F;textbooks&#x2F;approved-textbooks&#x2F;" rel="nofollow">https:&#x2F;&#x2F;aimath.org&#x2F;textbooks&#x2F;approved-textbooks&#x2F;</a><p>There&#x27;s also a TON of high-quality maths instructional content on Youtube, Videolectures.net, etc. For example, there&#x27;s some really good stuff by David McKay (also mentioned in CuriouslyC&#x27;s post) here:<p><a href="http:&#x2F;&#x2F;videolectures.net&#x2F;david_mackay&#x2F;" rel="nofollow">http:&#x2F;&#x2F;videolectures.net&#x2F;david_mackay&#x2F;</a><p>Be sure to check out Professor Leonard:<p><a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;user&#x2F;professorleonard57" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;user&#x2F;professorleonard57</a><p>Gilbert Strang:<p><a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;results?search_query=gilbert+strang" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;results?search_query=gilbert+strang</a><p>and 3blue1brown:<p><a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;channel&#x2F;UCYO_jab_esuFRV4b17AJtAw" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;channel&#x2F;UCYO_jab_esuFRV4b17AJtAw</a><p>as well.
评论 #15121229 未加载
评论 #15118816 未加载
WhitneyLand超过 7 年前
Surprising level of disagreement here on a few items for a sub field that has its own degree tracks.<p>Multiavariable calc you either &quot;abolsutely&quot; need or don&#x27;t really need. Should be well versed in graph theory, or don&#x27;t need it much.<p>Surely some of the contradiction is caused by different assumptions of what the goal is. But some of its hard to relate to as a reader. For example, I haven&#x27;t been in the field but but have tried to read enough to understand the concepts, and having studied graph theory I don&#x27;t see how it&#x27;s a top 5 recommendation.<p>I don&#x27;t doubt anyone&#x27;s experience, would just be nice to know which assumption is behind a suggestion.
评论 #15118422 未加载
sunsu超过 7 年前
None. You can be a productive ML engineer without understanding the math. Many elitist engineers here will downvote me, but its true. ML libraries that allow you to quickly get productive have come a long way. BUT, you have to have a solid understanding of WHICH algorithms&#x2F;tools to use WHEN. There is also a lot of &quot;voodoo&quot; knowledge to gain that isn&#x27;t well documented or explained (unrelated to maths).
gtani超过 7 年前
For going all in, <a href="https:&#x2F;&#x2F;ocw.mit.edu&#x2F;courses&#x2F;mathematics&#x2F;18-657-mathematics-of-machine-learning-fall-2015&#x2F;syllabus&#x2F;" rel="nofollow">https:&#x2F;&#x2F;ocw.mit.edu&#x2F;courses&#x2F;mathematics&#x2F;18-657-mathematics-o...</a><p>But a good number of people that are doing work haven&#x27;t taken real analysis, or it&#x27;s been awhile and so you should be current on multivariable and vector calculus. Calculus of variations shows up from time to time.<p>For math reviews, look at the following (there&#x27;s others if you want more refs, ping me):<p><a href="http:&#x2F;&#x2F;www.deeplearningbook.org&#x2F;" rel="nofollow">http:&#x2F;&#x2F;www.deeplearningbook.org&#x2F;</a><p><a href="https:&#x2F;&#x2F;metacademy.org&#x2F;roadmaps&#x2F;" rel="nofollow">https:&#x2F;&#x2F;metacademy.org&#x2F;roadmaps&#x2F;</a><p><a href="http:&#x2F;&#x2F;www.cs.huji.ac.il&#x2F;~shais&#x2F;UnderstandingMachineLearning&#x2F;understanding-machine-learning-theory-algorithms.pdf" rel="nofollow">http:&#x2F;&#x2F;www.cs.huji.ac.il&#x2F;~shais&#x2F;UnderstandingMachineLearning...</a>
ratsimihah超过 7 年前
Make sure to differentiate between AI researcher and applied AI software engineer, or whatever that is called.<p>The former needs the mathematical background mentioned here to develop groundbreaking algorithms or improve on existing ones, while the latter merely implements them and requires a much smaller mathematical toolset.
Myrmornis超过 7 年前
It depends whether you want to work more as an engineer &#x2F; data analyst, or more as a &quot;ML researcher&quot;. For the latter, then, yes, as everyone says below, you need to be totally comfortable with multivariable calculus, linear algebra, probability and statistics, numerical optimization etc. But many jobs are more practical in nature, in which the main case essential skill is, being able to run a bunch of different models with different parameter values and collect and interpret the results, efficiently and reproducibly, and be able to talk about them and make recommendations for the way forwards. In those jobs you&#x27;re not actually going to need to be able to derive updates for backpropagation, even though it&#x27;s certainly satisfying to understand it.
评论 #15117246 未加载
mtzet超过 7 年前
Honestly, I don&#x27;t think having to learn some stuff before starting anything is nessecary, especially for learning a field as wide as ML&#x2F;AI. It&#x27;s much better to start out trying to learning something you&#x27;re interested in, and then trying to fill in the gaps. This will also help you understand and motivate the underlying theory you&#x27;re reading.<p>So for example, start with some source in ML&#x2F;AI you&#x27;d like to read. If you get stuck, ask somewhere (possibly an online forum like this) what field you&#x27;re having trouble with and how to get started there.
wadams19超过 7 年前
Totally depends on where you want to land on the engineering-AI-products to pure-AI-research spectrum.<p>So, what do you mean by &quot;pursuing&quot;?<p>But even still, I would caution against trying to upload a bunch of new math concepts into your brain without first understanding the ML&#x2F;AI context.<p>I would say go through both of Andrew Ng&#x27;s ML and DL courses on Coursera.<p>Then, pick a domain&#x2F; problem that you&#x27;re interested in.<p>Then, read papers about how ML&#x2F;AI is applied in that domain.<p>Then, try to reproduce a paper that you understand and are interested in.
charlescearl超过 7 年前
Michael I. Jordan&#x27;s suggested reading list has been posted here a few times<p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=1055389" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=1055389</a>
coconut_crab超过 7 年前
Maybe unrelated to OP&#x27;s question but I have always felt that it is impossible to get a job in AI&#x2F;ML without a PhD in that field (by getting a job I mean do something new&#x2F;useful and not just coding algorithm devised by other people). I studied mechatronics in university and fairly comfortable with math (calculus, linear algebra and stats), I have even written a small neural network back then to optimise parameters for lathe machining. But that&#x27;s no where near enough for a job in AI&#x2F;ML. Unlike writing a web page, which someone can learn within a week to produce something usable, I feel like you need years and years of studying to barely get a start in ML&#x2F;AI and there is no hope for us non computer scientist at all.<p>[Added] Of course writing webpages pays well enough, but I still can&#x27;t shake off this feeling that I am missing something by not jumping on the AI&#x2F;ML train though.
amrrs超过 7 年前
Statistics and Probability - For non-math background, Openintro.org with R and Sas lab is a good one. Khan academy videos on the same again makes a lot of concepts easier.<p><a href="http:&#x2F;&#x2F;www.r-bloggers.com&#x2F;in-depth-introduction-to-machine-learning-in-15-hours-of-expert-videos&#x2F;" rel="nofollow">http:&#x2F;&#x2F;www.r-bloggers.com&#x2F;in-depth-introduction-to-machine-l...</a> Introduction to Statistical Learning <a href="http:&#x2F;&#x2F;www-bcf.usc.edu&#x2F;~gareth&#x2F;ISL&#x2F;" rel="nofollow">http:&#x2F;&#x2F;www-bcf.usc.edu&#x2F;~gareth&#x2F;ISL&#x2F;</a> (Rob S and by Trevor H, Free I guess) for more in depth, Elements of Statistical Learning by the same.<p>Linear Algebra (Andrew Ng&#x27;s this part in Introduction to Machine Learning is a short and crisp one)<p>If you&#x27;re not scared by Derivatives, you can check them. But you can easily survive and even excel as a data scientist or ML practitioner with these.
sumitgt超过 7 年前
I won&#x27;t really comment about ML&#x2F;AI in general. But, if you specifically care about getting into Deep Learning, I would say only bother looking into: - Basic linear algebra and matrix algebra.<p>Since you would rely on frameworks like Tensorflow to handle figuring out the derivatives for you, you don&#x27;t really need to know much calculus. Just read up on what derivative of a function at a particular point signifies. This should give you enough intuition to understand things initially.<p>A skill that would really come in useful would be ability to look at a function and think how increasing&#x2F;decreasing one of the variables would affect it&#x27;s value. This would help develop intuition around a lot of concepts used in Deep Learning topologies.
rdrey超过 7 年前
I previously attempted Andrew Ng&#x27;s old course, but didn&#x27;t complete the tutorials. Now I would start in this order:<p>Watch the course.fast.ai lectures quickly, just to see a lot of practical ML&#x2F;AI applications. You&#x27;ll see how effective you can be just by knowing the tools with very little math background.<p>Next I&#x27;d look at the NEW Andrew Ng introduction on Coursera. It is much more approachable than his first course. You might still feel a little overwhelmed by a few equations, but then you&#x27;ll implement them yourself in numpy. (And the ipython&#x2F;jupyter notebooks are really well written, walking you through every step.)
mendeza超过 7 年前
I wish the people who answer this question are people that are current deep learning engineers or data scientist that use deep learning in real world settings, I am worried that people who are not credible are giving advice, which is not valuable. I am a masters student taking a PhD class in Bayesian machine learning to figure this out as well. I hope to have a better answer for this by the end of the course!
评论 #15118717 未加载
评论 #15118334 未加载
jochenleidner超过 7 年前
1. You can get a long way with high school calculus and probability theory.<p>2. Regarding books I second the late David McKay&#x27;s &quot;Information Theory, Inference and Learning Algorithms&quot; and the second edition of &quot;Elements of Statistical Learning&quot; by Tibshirani et al. (there&#x27;s also a more accessible version of a subset of the material targeting MBA students called James et al., An Introduction to Statistical Learning). Duda&#x2F;Hart&#x2F;Stork&#x27;s Pattern Classification (2nd ed.) is also great. The self-published volume by Abu-Mostafa&#x2F;Magdon-Ismail&#x2F;Lin, Learning from Data: A Short Course is impressive, short and useful for self-study.<p>3. Wikipedia is surprisingly good at providing help, and so is Stack Exchange, which has a statistics sub-forum, and of course there are many online MOOC courses on statistics&#x2F;probability and more specialized ones on machine learning.<p>4. After that you will want to consult conference papers and online tutorials on particular models (k-means, Ward&#x2F;HAC, HMM, SVM, perceptron, MLP, linear and logistic regression, kNN, multinomial naive Bayes, ...).
xchip超过 7 年前
All I needed to write my conv net library was to understand the chain rule and some basic multiplication.<p>People like to make this look harder than it is.
ivan_ah超过 7 年前
Probability theory and linear algebra are pretty much the core. Learning LA will help you become comfortable with multi-dimensional quantities, vector spaces, and give you some powerful computational techniques, e.g., SVG==PCA.<p>Here is a short tutorial on linear algebra: <a href="https:&#x2F;&#x2F;minireference.com&#x2F;static&#x2F;tutorials&#x2F;linear_algebra_in_4_pages.pdf" rel="nofollow">https:&#x2F;&#x2F;minireference.com&#x2F;static&#x2F;tutorials&#x2F;linear_algebra_in...</a> and a preview of the full book: <a href="https:&#x2F;&#x2F;minireference.com&#x2F;static&#x2F;excerpts&#x2F;noBSguide2LA_preview.pdf" rel="nofollow">https:&#x2F;&#x2F;minireference.com&#x2F;static&#x2F;excerpts&#x2F;noBSguide2LA_previ...</a>
blt超过 7 年前
If you want to understand SVMs deeply, a course in convex optimization. In general, proving maximum likelihood estimation for a lot of classic machine learning models involves using the method of Lagrange multipliers. But not deep neural networks :)
pramalin超过 7 年前
I found this study plan very useful for me. <a href="https:&#x2F;&#x2F;www.analyticsvidhya.com&#x2F;blog&#x2F;2017&#x2F;01&#x2F;the-most-comprehensive-data-science-learning-plan-for-2017&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.analyticsvidhya.com&#x2F;blog&#x2F;2017&#x2F;01&#x2F;the-most-compre...</a><p>Provides a very good idea of the courses required and their time frame. I roughly followed along this path but took &quot;Analytics Edge&quot; <a href="https:&#x2F;&#x2F;www.edx.org&#x2F;course&#x2F;analytics-edge-mitx-15-071x-3" rel="nofollow">https:&#x2F;&#x2F;www.edx.org&#x2F;course&#x2F;analytics-edge-mitx-15-071x-3</a> for introduction into ML algorithms.
framebit超过 7 年前
A thorough, intuitive grounding in statistics is crucial, IMO.<p>Doing any kind of ML means questioning all the assumptions that go into your results and understanding how those assumptions could affect the outcome. That process starts in stats.
bjourne超过 7 年前
It depends on what &quot;pursuing ML&#x2F;AI&quot; means. I&#x27;ve written a recommendation engine with barely understanding linear algebra and a spam filter without knowing Bayes theorem. A programmer can work on ML systems without having a solid foundation in higher maths. However, if you want to develop your own solutions then you surely need the math.<p>I would recommend reading Toby Segaran&#x27;s Programming Collective Intelligence: <a href="http:&#x2F;&#x2F;shop.oreilly.com&#x2F;product&#x2F;9780596529321.do" rel="nofollow">http:&#x2F;&#x2F;shop.oreilly.com&#x2F;product&#x2F;9780596529321.do</a>
ChadyWady超过 7 年前
For ML, the other users gave a good coverage of topics. But AI is an incredibly broad field, and each specialty uses different math topics. Learning all of the math would be infeasible. What are your particular interests?<p>Russell and Norvig have a good book at <a href="http:&#x2F;&#x2F;aima.cs.berkeley.edu" rel="nofollow">http:&#x2F;&#x2F;aima.cs.berkeley.edu</a> that covers many different topics in AI, although it is definitely not comprehensive. I would say that whatever you learn in an undergraduate CS degree would give you a good starting point for learning any particular AI topics.
septimus111超过 7 年前
Not knowing anything about you, I&#x27;ll assume that<p>- you are starting with the equivalent of a high school level of maths<p>- you want to take a ML course or read an ML book without feeling totally lost<p>As some commenters have said, Calculus, Probability and Linear Algebra will be very helpful.<p>Some people like to recommend the &quot;best&quot; or &quot;most important&quot; books which you &quot;should&quot; read, but there is a strong chance these will end up sitting on a bookshelf, barely touched. So I will recommend some books which are perhaps more accessible.<p>- Calculus by Gilbert Strang<p>- Linear Algebra by Gilbert Strang<p>For Probability: I don&#x27;t have any favourites, sorry.
pramalin超过 7 年前
You can also dive in first and then cover the math behind ML, by taking Andrew Ng&#x27;s courses. <a href="https:&#x2F;&#x2F;www.coursera.org&#x2F;learn&#x2F;machine-learning" rel="nofollow">https:&#x2F;&#x2F;www.coursera.org&#x2F;learn&#x2F;machine-learning</a> <a href="https:&#x2F;&#x2F;www.coursera.org&#x2F;specializations&#x2F;deep-learning" rel="nofollow">https:&#x2F;&#x2F;www.coursera.org&#x2F;specializations&#x2F;deep-learning</a>
deepnotderp超过 7 年前
Basic high school calculus and linear algebra is really the only <i>required</i> thing.<p>I would recommend probability theory and statistics as well.
e19293001超过 7 年前
You can learn the required maths along the way through Andrew Ng&#x27;s deep learning course at coursera.
pveierland超过 7 年前
The following is a concise and good explanation of necessary knowledge of information theory:<p><a href="http:&#x2F;&#x2F;colah.github.io&#x2F;posts&#x2F;2015-09-Visual-Information&#x2F;" rel="nofollow">http:&#x2F;&#x2F;colah.github.io&#x2F;posts&#x2F;2015-09-Visual-Information&#x2F;</a>
jules超过 7 年前
* Calculus<p>* Linear algebra<p>* Optimisation<p>* Probability<p>Various universities have very good course content freely available online, often including textbook recommendations, course notes, exercises, sample exams, and video lectures. Realistically it is probably going to be quite difficult to learn this on your own.
ronald_raygun超过 7 年前
I&#x27;d say everything in this list is good to know <a href="http:&#x2F;&#x2F;pages.cs.wisc.edu&#x2F;~tdw&#x2F;files&#x2F;cookbook-en.pdf" rel="nofollow">http:&#x2F;&#x2F;pages.cs.wisc.edu&#x2F;~tdw&#x2F;files&#x2F;cookbook-en.pdf</a>
评论 #15125446 未加载
wickedgamer超过 7 年前
Calculus Functions Derivatives Integration Analytic Geometry That&#x27;s all I think *<a href="http:&#x2F;&#x2F;shrugemojis.com&#x2F;shrug-emoji&#x2F;" rel="nofollow">http:&#x2F;&#x2F;shrugemojis.com&#x2F;shrug-emoji&#x2F;</a>
dtjon超过 7 年前
Probability, and thus multivariate calculus and partial differential equations. Linear algebra. Convex Optimization, and thus multivariate and partial differential equations. Some principals of statistics is usually helpful
评论 #15117338 未加载
EternalData超过 7 年前
Some people have had a more comprehensive view on this -- if I were to focus on one field of math to understand really well though, it&#x27;d be statistical reasoning and the understanding of probability and uncertainty.
bitL超过 7 年前
Calculus (preferably both multi-variate and discrete), probability, statistics, operations research, graph theory, topology, computational complexity. All depends on how deep you&#x27;d like to go.
评论 #15120709 未加载
wickedgamer超过 7 年前
Yet it depends. Theres a lot out there on google one could learn. <a href="http:&#x2F;&#x2F;fitnessjab.com&#x2F;" rel="nofollow">http:&#x2F;&#x2F;fitnessjab.com&#x2F;</a>
master_yoda_1超过 7 年前
There are two way to approach ML&#x2F;AI<p>1) First read all the prerequisites and then work on a problem<p>2) Start working on a problem and learn all the math ML&#x2F;AI as you need<p>The second option works best.
graycat超过 7 年前
Part I<p>(1) Calculus<p>Generally should have college freshman and sophomore calculus.<p>(1.1) Functions<p>So, there can understand better what a <i>function</i> is. E.g., function<p><pre><code> f(x) = 3x^2 + 1. </code></pre> (1.2) Derivatives<p>Then will learn how to find the slope of the graph of a function. That is the <i>derivative</i> of the function. E.g., for function f with f(x) = 3x + 2, as in high school algebra, the slope is 3. Then for each x, the derivative of f at x is just 3.<p>The derivative of function f is denoted by either of<p><pre><code> f&#x27;(x) = d&#x2F;dx f(x) </code></pre> E.g., for function f(x) = 3x^2 + 1 it turns out that<p><pre><code> f&#x27;(x) = 6x. </code></pre> (1.3) Integration<p>For function<p><pre><code> g(x) = 6 x </code></pre> maybe we want to know what function f(x) will give us<p><pre><code> f&#x27;(x) = g(x) </code></pre> Finding such a function f is <i>anti-differentiation</i>, that is, undoes differentiation. So, sure,<p><pre><code> f(x) = 3x^2 + C </code></pre> for any constant C.<p>Such anti-differentiation is also the way to find the area under a curve. So, can use that to find the area of a circle, volume of a cylinder, etc. Doing that the anti-differentiation is <i>integration</i>.<p>The fundamental theorem of calculus shows how differentiation and integration are related.<p>(1.4) Analytic Geometry<p>Commonly taught at the beginning of a calculus course is <i>analytic</i> geometry.<p>So, take a cone an cut it. Then the cut surfaces will be one of a circle, an ellipse, a parabola, a hyperbola, or just two crossed straight lines. So, those curves are from a cone and are the <i>conic sections</i>.<p>There is some simple associated algebra.<p>Conic sections are important off and on; e.g., applied math is awash in circles; the planets move in ellipses; a baseball moves in a parabola or nearly so; an electron moving toward a negative charge will turn away from that charge in a hyperbola.<p>It turns out that in linear algebra (below) circles and ellipses are important.<p>(1.5) Role of Calculus<p>Calculus was invented by Newton as part of working with force and acceleration for understanding the motion of the planets.<p>E.g., if at time t function d(t) gives distance traveled, then function v(t) = d&#x27;(t) is the velocity at time t and function a(t) = v&#x27;(t) is the acceleration at time t.<p>Then Newton&#x27;s second law is<p><pre><code> F(t) = m a(t) </code></pre> where F(t) is the force at time t applied to mass m.<p>Calculus is the first approach to the analysis of continuous change and is a pillar of civilization.<p>Knowledge of calculus will commonly be assumed in work in ML&#x2F;AL, data science, statistics, optimization, applied math, engineering, etc.<p>E.g., a lot in ML, AI, and data science is getting best fits to data; best fitting is to minimize errors in the fit; such minimization is mostly a calculus problem; one of the main steps in ML is steepest descent, and that is from a derivative.<p>Probability theory (e.g., evaluating coin tossing, poker hands, accuracy in ML) will be important in ML&#x2F;AI, etc.; two of the basic notions in probability are cumulative distributions and density distributions; the cumulative is from an integration, and the density is from a differentiation.
评论 #15118409 未加载
bootcat超过 7 年前
Probability and Statistics to begin with !
bluetwo超过 7 年前
Not a mention so far about game theory or Nash equilibrium.<p>I&#x27;m no expert but does anyone think these apply?
评论 #15118726 未加载
评论 #15118211 未加载
dekhn超过 7 年前
Linear algebra, probability, and tree&#x2F;graphs.
mindhash超过 7 年前
demystified has good series on calculus and linear algebra. Its light weight
blubb-fish超过 7 年前
just start learning and you will see what math comes along ;)
adamnemecek超过 7 年前
Stanford EE263 is very spicy <a href="http:&#x2F;&#x2F;ee263.stanford.edu" rel="nofollow">http:&#x2F;&#x2F;ee263.stanford.edu</a>
proofofstake超过 7 年前
&gt; What maths must be understood to enable pursuit of either of the above fields?<p>None.<p>&gt; Are there any seminal texts&#x2F;courses&#x2F;content which should be consumed before starting?<p>No.<p>You don&#x27;t need to know binary to start being a programmer&#x2F;developer either. Just start already. As long as you are not in charge of a medical diagnosis or financial model, you don&#x27;t get any drawback in experimenting (and failing miserably).<p>Assuming applied ML, the most difficult part will be the human-political business element of it: People not understanding your model or using its output correctly, bias, feedback loops, acquiring enough resources, etc. The more you can explain to them, without resorting to heavy maths, the better communicator you are.<p>That said, it can&#x27;t hurt to do Ng&#x27;s Coursera course (a lot of top performers started out with this course). Learning from Data by Caltech&#x27;s Abu-Mostafa goes very wide on machine learning. &quot;Programming Collective Intelligence&quot; is a, somewhat dated, good book.<p>As for seminal texts, the field is too wide for this. A better bet is: Find a professor in the field you are interested in. Say &quot;Deep Learning&quot;, you could have a look at LeCun, Hinton, Schmidhuber, Bengio, ... Now look at their PhD-students, their papers, their courses, their conference talks, their software, their current research. Basically become a student under the most authoritative professor in the subfield you can find and resonate with, without ever paying any university tuition or them knowing you exist. This is very possible these days.<p>But by all means: Just start out. Machine learning is fun. Learning about dry 100 year old maths not so much. Make mistakes. Learn to detect and avoid overfit. Find out if you are passionate and curious about parts of the field, then the theory will come eventually. A lot of the time these questions seem to demand answers like: &quot;You need a PhD-level understanding of mathematics&quot; Just so your brain can go: &quot;I am not good enough for this, so let&#x27;s look at something easier&quot;. Don&#x27;t use this as an excuse. Start making intelligent stuff. There are 16-year-olds on Kaggle routinely beating maths PhD&#x27;s.<p>Also remember that, despite the current trend of calling everything &quot;AI&quot;, that AI is a very wide field, of which mathematics is only a small part. There is philosophy, linguistics, cognitive science, physics, neuroscience, psychology, computer science, robotics, logic, ... all these parts vary wildly in their prerequisite maths knowledge.