Archive

Posts Tagged ‘Education’

Open Course Website & Podcast

April 22, 2012 Leave a comment

一、伯克利

加州大学伯克利分校 http://webcast.berkeley.edu/courses.php

作为美国第一的公立大学,伯克利分校提供了,可以跟踪最新的讲座。想看教授布置的作业和课堂笔记,可以点击该教授的网页,通常,他/她都会第一堂课留下网址。实在不行,用google搜搜吧!

  伯克利的视频都是.rm格式,请注意转换

  二、麻省

麻省理工学院 http://ocw.mit.edu/OcwWeb/web/courses/courses/index.htm

麻省理工是免费开放教育课件的先驱,计划在今年把1800门课程的课件都放在网站上,提供课程与作业的PDF格式下载。三是,麻省理工只提供少数的视频讲座。坐过学生上麻省有一个绝对优势,麻省理工在中国大陆和中国台湾都建立了镜像网站,把麻省的课程都翻译成立中文。鉴于PDF格式,推荐使用FoxIt Reader。

  www.core.org.cn(中国大陆)推荐

  www.myoops.org(中国台湾)

二、卡耐基梅隆 http://www.cmu.edu/oli/

卡耐基梅隆针对初入大学的大学生,提供10门学科的课程视频。与其他大学的免费课程一样,非卡耐基梅隆的学子能学习课程,但是为了使学生能够及时了解自己的课程进度,卡耐基梅隆建议造访者在网站上注册,建立自己的资料库。这样一来,你得在有限的时间内完成一门课程,还要参加几次考试,当然,即使你得了100分,卡耐基梅隆也不会给你开证明,更不会给你学分。

 

四、犹他

犹他大学 http://ocw.usu.edu/front-page/Courese_listing

犹他大学类似于麻省理工,提供大量的课程课件

 

五、塔夫茨

塔夫茨大学 http://ocw.tufts.edu

塔夫茨大学也是“开放式教育课程”的先驱之一,初期提供的课程着重在本校专长的生命科学、跨领域方法、国际观点以及对美国地区性、全国性社群服务的基础理论。

 

六、公开

英国公开大学 http://openlearn.open.ac.uk/course/index.php

英国十几所大学联合起来,组建了英国公开大学。有一部分课程是对注册学生开放的,但是有一批很好的课程是免费的,并提供视频。每门课还设立了论坛,在社区中,大家发表意见,提供其他的学习资源,互相取经。在这个网站里,最能锻炼自学者的能力,因为你要不停地淘,才能找到宝贝。

 

七、约翰霍普金斯

约翰霍普金斯 http://ocw.jhsph.edu/topics.cfm

只有极少数人能够进入约翰霍普金斯大学就读,但是,现如今有动机的人不用花一分钱,便能通过网站获得该校的前沿知识。约翰霍普金斯提供了本学院最受欢迎的课程,包括青少年健康、行为和健康、生物统计学、环境、一般公共卫生、卫生政策、预防伤害、母亲和儿童健康、心理卫生、营养、人口科学、公共卫生准备和难民卫生等。

 

八、Connexions

http://cnx.rice.edu

CNX.org由莱斯大学开发,号称是课程资源免费共享图书馆。与其他大学不同的事,CNX邀请教授学者建立自己的社区,把自己的最新成果公布于世,接受大家的评价。可以说CNX开辟了大学资源共享的新天地,尤其适合自学能力超强的大学生。有些课程有中文版。

 

九、索菲亚

索菲亚大学 http://sofia.ocw.cn/gallery

无论是想当一名管理者、作家、评论员、还是要从事设计和IT业的人,索菲亚大学的免费课程肯定让你受益匪浅。索菲亚大学提供了8门学科的课程,其中《企业网络安全实战》已翻译成中文。

 

十、华盛顿

华盛顿大学 http://www.cs.washington.edu/education/course-webs.html

华盛顿大学的计算机工程学比较强,相关的几百门课程都已经放到网上。不但本科生能找到所需要的课程,连研究生也能淘到宝贝。该网站还提供特色讲座,比如:妇女、计算机与合作。课程不但提供讲座介绍、课堂笔记、有些课程还提供视频。

 

牛津、斯坦福、耶鲁大学联合网站

http://www.alllearn.org

哥伦比亚大学

http://ci.columbia.edu/ci

伯克利音乐学院

http://www.berkleeshares.com

杜克大学法律中心

http://www.law.duke.edu/cspd/lectures

圣母大学

http://ocw.nd.edu

英国格雷莎姆学院

http://www.gresham.ac.uk/default.asp

加州大学Irvine分校

http://ocw.uci.edu

富布莱特学校

http://ocw.fetp.edu.vn/home.cfm

日本东京大学

http://ocw.u-tokyo.ac.jp/english

日本早稻田大学

http://www.waseda.jp/ocw/index.html

日本大阪大学

http://ocw.osaka-u.ac.jp/index.php

法国巴黎高科

http://graduateschool.paristech.org

 

 

再给个看大学视频播客的

 

斯坦福大学 http://itunes.stanford.edu

加州大学伯克利分校http://itunes.berkeley.edu

普渡大学http://boilercast.itap.purdue.edu:1013/Boilercast

美国西南理工http://pocast.swtc.edu/lecture/index.php

加州大学洛杉矶分校http://www.bruincast.ucla.edu

西肯塔基大学http://blog.wku.edu/podcasts

Dupage学院http://www.cod.edu/multimedia/podcast/CODcast/Welcome.html

纽约城市大学http://podcast.york.cuny.edu/lectures

莱斯大学http://webcast.rice.edu

加州大学圣地亚哥分校http://podcast.ucsd.edu

剑桥大学http://mediaplayer.group.cam.ac.uk/main/Podcasts.html

美国大学华盛顿法律学院 http://www.wcl.american.edu/podcasts

杜克大学法律学院http://www.law.duke.edu/webcast

乔治敦大学 http://webcast.georgetown.edu

芝加哥大学商学院 http://www.chicagogsb.edu/multimedia/podcast

波士顿学院 http://frontrow.bc.edu

哈佛商学院在线

http://www.hbsp.harbard.edu/b02/en/hbr_ideacast.jhtml;jsessioned=NVHF0YFBS5ZCGAKRGWDR5VQBKE0YIISW

威斯康星麦迪逊大学 http://havenscenter.org/audio/audio.htm

约翰霍普金斯大学 http://www.johnshopkins.edu/podcasts.index1.html

伦敦政治经济学院 http://www.lse.ac.uk/rescources/podcasts/Default.htm

普林斯顿大学 http://uc.princeton.edu/main/index.php

英国泰晤士报MBA http://uc.princeton.edu/main/index.php

耶鲁大学 http://www.yale.edu/opa/podcast/

Advertisements
Categories: Education Tags:

Distributed Learning: A new model

April 19, 2012 Leave a comment

The Geomblog

via Distributed Learning: A new model.

Communication is now the key to modelling distributed/multicore computations. Jim Demmel has been writing papers and giving talks on this theme for a while now, and as processors get faster, and the cloud becomes a standard computing platform, communication between nodes is turning out to be the major bottleneck.

So suppose you want to learn in this setting ? Suppose you have data sitting on different nodes (you have a data center, or a heterogeneous sensor network, and so on) and you’d like to learn something on the union of the data sets. You can’t afford to ship everything to a single server for processing: the data might be too large to store, and the time to ship might be prohibitive. 

So can you learn over the (implicit) union of all the data, with as little discussion among nodes as possible ? This was the topic of my Shonan talk, as well as two papers that I’ve been working on with my student Avishek Saha, in collaboration with  Jeff Phillips and Hal Daume. The first one will be presented at AISTATSthis week, and the second was just posted to the arxiv.

We started out with the simplest of learning problems: classification. Supppose you have data sitting on two nodes (A and B), and you wish to learn a hypothesis over the union of A and B. What you’d like is a way for the nodes to communicate as little as possible with each other while still generating a hypothesis close to the optimal solution.

It’s not hard to see that you could compute an $\epsilon$-sample on A, and ship it over to B. By the usual properties of an $\epsilon$-sample, you guarantee that any classifier on B’s data combined with the sample will also classify A correctly to within some $\epsilon$-error. It’s also not too hard to show a lower bound that matches this upper bound. The amount of communication is nearly linear in $1/\epsilon$. 

But can you do better ? In fact yes, if you let the nodes talk to each other, rather than only allowing one-way communication. One way of gaining intuition for this is that $A$ can generate classifiers, and send them over to $B$, and $B$ can tell $A$ to turn the classifier left or right. Effectively, $B$ acts as an oracle for binary search. The hard part is showing that this is actually a decimation (in that a constant fraction of points are eliminated from consideration as support points in each step), and once we do that, we can show an exponential improvement over one-way communication. There’s a trivial way to extend this to more than 2 players, with a $k^2$ blow up in communication for $k$ players. 

This binary search intuition only works for points in 2D, because then the search space of classifiers is on the circle, which lends itself naturally to a binary search. In higher dimensions, we have to use what is essentially a generalization of binary search – the multiplicative weight update method. I’ll have more to say about this in a later post, but you can think of the MWU as a “confused zombie” binary search, in that you only sort of know “which way to go” when doing the search, and even then points that you dismissed earlier might rise from the dead. 

It takes a little more work to bring the overhead for k-players down to a factor k. This comes by selecting one node as a coordinator, and implementing one of the distributed continuous sampling techniques to pass data to the coordinator. 

You can read the paper for more details on the method. One thing to note is that the MWU can be “imported” from other methods that use it, which means that we get distributed algorithms for many optimization problems for free. This is great because a number of ML problems essentially reduce to some kind of optimization. 

A second design template is multipass streaming: it’s fairly easy to see that any multipass sublinear streaming algorithm can be placed in the k-player distributed setting, and so if you want a distributed algorithm, design a multipass streaming algorithm first. 

One weakness of our algorithms was that we didn’t work in the “agnostic” case, where the optimal solution itself might not be a perfect classifier (or where the data isn’t separable, to view it differently). This can be fixed: in an arxiv upload made simultaneously with ours, Blum, Balcan, Fine and Mansour solve this problem very neatly, in addition to proving a number of PAC-learning results in this model.

It’s nice to see different groups exploring this view of distributed learning. It shows that the model itself has legs. There are a number of problems that remain to be explored, and I’m hoping we can crack some of them. In all of this, the key is to get from a ‘near linear in error’ bound to a ‘logarithmic in error’ bound via replacing sampling by active sampling (or binary search). 

%d bloggers like this: